code/r.ipynb

{
  "cells": [
    {
      "id": "a8e4f386",
      "cell_type": "markdown",
      "source": "[\u8fbb\u771f\u543e\u30fb\u77e2\u5439\u592a\u6717\u300e\u30bc\u30ed\u304b\u3089\u306f\u3058\u3081\u308b\u30c7\u30fc\u30bf\u30b5\u30a4\u30a8\u30f3\u30b9\u5165\u9580\u300f\uff08\u8b1b\u8ac7\u793e,\u00a02021\uff09](https://github.com/taroyabuki/fromzero)",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u3053\u308c\u306fR\u306e\u30b3\u30fc\u30c9\u306e\u4f8b\u3067\u3059\uff0e\n",
        "1 + 1"
      ],
      "id": "240b5c18-7d52-4447-963e-2757f7585738"
    },
    {
      "id": "421a31bd",
      "cell_type": "markdown",
      "source": "# 1 \u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u3068\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\n\n\n",
      "metadata": {}
    },
    {
      "id": "215431b9",
      "cell_type": "markdown",
      "source": "## 1.1 \u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u306e\u57fa\u672c\u64cd\u4f5c\n\n\n",
      "metadata": {}
    },
    {
      "id": "7baa80d1",
      "cell_type": "markdown",
      "source": "## 1.2 \u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u306e\u4ed5\u7d44\u307f\n\n\n",
      "metadata": {}
    },
    {
      "id": "1a7a0351",
      "cell_type": "markdown",
      "source": "# 2 \u30c7\u30fc\u30bf\u30b5\u30a4\u30a8\u30f3\u30b9\u306e\u305f\u3081\u306e\u74b0\u5883\n\n\n",
      "metadata": {}
    },
    {
      "id": "d4299150",
      "cell_type": "markdown",
      "source": "## 2.1 \u5b9f\u884c\u74b0\u5883\u306e\u9078\u629e\n\n\n",
      "metadata": {}
    },
    {
      "id": "231a0999",
      "cell_type": "markdown",
      "source": "## 2.2 \u30af\u30e9\u30a6\u30c9\n\n\n",
      "metadata": {}
    },
    {
      "id": "3373d73e",
      "cell_type": "markdown",
      "source": "## 2.3 Docker\n\n\n",
      "metadata": {}
    },
    {
      "id": "250f557a",
      "cell_type": "markdown",
      "source": "## 2.4 \u30bf\u30fc\u30df\u30ca\u30eb\u306e\u4f7f\u3044\u65b9\n\n\n",
      "metadata": {}
    },
    {
      "id": "7d2bc194",
      "cell_type": "markdown",
      "source": "## 2.5 R\u3068Python\n\n\n",
      "metadata": {}
    },
    {
      "id": "c4bb5a45",
      "cell_type": "markdown",
      "source": "## 2.6 \u30b5\u30f3\u30d7\u30eb\u30b3\u30fc\u30c9\u306e\u5229\u7528",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "1 + 1\n",
        "\n",
        "print(1 + 2)\n",
        "\n",
        "1 + 3"
      ],
      "id": "ff9d95bc-b223-452f-b594-987f17a9d55a"
    },
    {
      "id": "0fb9d3df",
      "cell_type": "markdown",
      "source": "# 3 R\u3068Python",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Google Colaboratory\u306e\u74b0\u5883\u8a2d\u5b9a\n",
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  options(Ncpus = parallel::detectCores())\n",
        "  installed_packages <- rownames(installed.packages())\n",
        "  packages_to_install <- c(\"furrr\", \"proxy\", \"zeallot\")\n",
        "  install.packages(setdiff(packages_to_install, installed_packages))\n",
        "  install.packages(c(\"ggplot2\"))\n",
        "}"
      ],
      "id": "e0c0235b-3aff-4668-ac7a-2344f36a29ed"
    },
    {
      "id": "9d5b5352",
      "cell_type": "markdown",
      "source": "## 3.1 \u5165\u9580",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "0x10"
      ],
      "id": "7cca6403-6a75-41aa-9e22-fec1ec513779"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "1.23e5"
      ],
      "id": "1475b9bb-db85-4945-95d7-ebcc33434a25"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "2 * 3"
      ],
      "id": "3178e52b-c975-4079-81e7-74ea791bba29"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "10 / 3"
      ],
      "id": "cafff2a4-b309-49d1-a8bc-2d8dd378da7d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "10 %/% 3 # \u5546\n",
        "\n",
        "10 %% 3  # \u4f59\u308a"
      ],
      "id": "ba32139b-292e-4411-bda4-887ad8889ae7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- 2\n",
        "y <- 3\n",
        "x * y\n",
        "\n",
        "library(zeallot)\n",
        "c(x, y) %<-% c(20, 30) # \u307e\u3068\u3081\u3066\u540d\u4ed8\u3051\n",
        "x * y"
      ],
      "id": "b8b5eeb8-2b6f-4fd9-bcb1-6e79b7728891"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- 1 + 1\n",
        "# \u3053\u306e\u6bb5\u968e\u3067\u306f\u7d50\u679c\u306f\u8868\u793a\u3055\u308c\u306a\u3044\n",
        "\n",
        "x # \u5909\u6570\u540d\u3092\u8a55\u4fa1\u3059\u308b\uff0e"
      ],
      "id": "ec566762-762f-466c-8856-20f4065b98cf"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_s <- \"abcde\""
      ],
      "id": "879763eb-4668-4cfa-8665-08544e5b269c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "nchar(my_s)"
      ],
      "id": "63b203b6-cd2c-475c-b2ae-2f9ef3f2a24e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "str_c(\"This is \", \"a\", \" pen.\")"
      ],
      "id": "dab73eb0-9083-4398-8649-a38cecd2897b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "substr(x = my_s, start = 2, stop = 4)"
      ],
      "id": "99411aac-eecc-469d-b13f-e2b5d545f2e9"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- \"%s is %s.\"\n",
        "sprintf(tmp, \"This\", \"a pen\")"
      ],
      "id": "7b6b127c-f9db-4d9b-8e50-c419b2fe21c4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "1 <= 2\n",
        "\n",
        "1 < 0"
      ],
      "id": "ff6a38c9-d822-4990-951b-be028878874b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "0.1 + 0.1 + 0.1 == 0.3\n",
        "\n",
        "all.equal(0.1 + 0.1 + 0.1, 0.3)"
      ],
      "id": "1196e3d5-7cc9-4a3b-a2a2-fa83f5c80229"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "TRUE & FALSE # \u8ad6\u7406\u7a4d\uff08\u304b\u3064\uff09\n",
        "\n",
        "TRUE | FALSE # \u8ad6\u7406\u548c\uff08\u307e\u305f\u306f\uff09\n",
        "\n",
        "!TRUE        # \u5426\u5b9a\uff08\u3067\u306a\u3044\uff09"
      ],
      "id": "9a3ee5b5-1417-4b33-bd56-21374c55031e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "ifelse(3 < 5, 0, 10)"
      ],
      "id": "75b5ad73-65d1-42a0-b053-bb085a6fcf42"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "getwd()"
      ],
      "id": "bd00f66f-7232-4156-927e-071d548295ca"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "setwd(\"..\")\n",
        "getwd()"
      ],
      "id": "7ceff68d-9a5d-4e5b-8350-ef9f1049daf5"
    },
    {
      "id": "fb92e107",
      "cell_type": "markdown",
      "source": "## 3.2 \u95a2\u6570",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "sqrt(4)"
      ],
      "id": "3c681445-937e-4719-987b-6ba5f42003d0"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "log(100, 10)"
      ],
      "id": "a60939ef-6d03-4715-9e05-4b4d3fdb7456"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "log(100)         # \u81ea\u7136\u5bfe\u6570\n",
        "# \u3042\u308b\u3044\u306f\n",
        "log(100, exp(1)) # \u7701\u7565\u3057\u306a\u3044\u5834\u5408\n"
      ],
      "id": "4bf60531-c9f5-4071-88d0-f66109504d9f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "log10(100) # \u5e38\u7528\u5bfe\u6570\n",
        "\n",
        "log2(1024) # \u5e95\u304c2\u306e\u5bfe\u6570"
      ],
      "id": "bf19af0a-166d-4e17-a559-ee532d8741bf"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "4 %>% sqrt"
      ],
      "id": "74342ce9-8a39-4244-b1d4-81643334bd16"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "exp(log(5))       # \u901a\u5e38\u306e\u66f8\u304d\u65b9\n",
        "# \u3042\u308b\u3044\u306f\n",
        "5 %>% log %>% exp # \u30d1\u30a4\u30d7\u3092\u4f7f\u3046\u66f8\u304d\u65b9\n"
      ],
      "id": "7c1ae7cf-90ba-4c65-a8f8-32e76695cc35"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "f <- function(a, b) {\n",
        "  a - b\n",
        "}"
      ],
      "id": "082db692-a091-415d-a01b-447995bb6bbe"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "f(3, 5)"
      ],
      "id": "90c703a1-4291-42bd-9a09-03ef2282af8f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "f <- function(a, b = 5) {\n",
        "  a - b\n",
        "}\n",
        "\n",
        "f(3) # f(3, 5)\u3068\u540c\u3058\u3053\u3068"
      ],
      "id": "ba83a6af-6bc5-4470-a155-8145bb34c147"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "(function(a, b) { a - b })(3, 5)"
      ],
      "id": "0d21f7f9-bb6d-4aa1-9567-9371740a4564"
    },
    {
      "id": "ebdcd65e",
      "cell_type": "markdown",
      "source": "## 3.3 \u30b3\u30ec\u30af\u30b7\u30e7\u30f3",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- c(\"foo\", \"bar\", \"baz\")"
      ],
      "id": "ac675c4f-f49c-4e07-a445-dec46fdf2d55"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "length(x)"
      ],
      "id": "08c706a9-bb33-4158-a814-337e33f88bcc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x[2]"
      ],
      "id": "bc21529b-9da5-4be6-a067-0fd993909a86"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x[2] <- \"BAR\"\n",
        "x # \u7d50\u679c\u306e\u78ba\u8a8d\n",
        "\n",
        "x[2] <- \"bar\" # \u5143\u306b\u623b\u3059\uff0e"
      ],
      "id": "a21fc02a-a231-439b-955a-ef7c5dbc954e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x[-2]"
      ],
      "id": "626ecf42-03e1-46fe-94cc-9da913cff8d6"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "c(x, \"qux\")"
      ],
      "id": "cd276b4b-03ff-4bec-b6e2-a83b5e795558"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- c(x, \"qux\")\n",
        "x # \u7d50\u679c\u306e\u78ba\u8a8d"
      ],
      "id": "db6a1dcc-b96b-49a0-b537-ac44b4062d59"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "1:5"
      ],
      "id": "e987374b-e68a-43f5-935e-035e49f5eb7f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "seq(from = 0, to = 10, by = 2)"
      ],
      "id": "1fc65ac4-907b-4d70-b4e6-1879e586e215"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "seq(from = 0, to = 1, by = 0.5)"
      ],
      "id": "c28667a8-c011-4685-934b-8eabcaa71c29"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "seq(from = 0, to = 100, length.out = 5)"
      ],
      "id": "a50bdb95-7e17-4dd4-8b59-a8252ad07501"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "rep(x = 10, times = 5)"
      ],
      "id": "5ca0deae-88ab-4405-a6f4-74b389b2fe28"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- c(\"\u30b0\u30fc\", \"\u30d1\u30fc\", \"\u30b0\u30fc\", \"\u30d1\u30fc\")\n",
        "x <- factor(tmp, levels = c(\"\u30b0\u30fc\", \"\u30c1\u30e7\u30ad\", \"\u30d1\u30fc\"))\n",
        "x"
      ],
      "id": "2adbb709-d924-4e18-8912-dd503082aa4e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- c(2, 3, 5, 7)\n",
        "\n",
        "x + 10 # \u52a0\u7b97\n",
        "\n",
        "x * 10 # \u4e57\u7b97"
      ],
      "id": "b7d17fd8-f47f-427a-bf9e-2843b095843e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- c(2, 3)\n",
        "sin(x)"
      ],
      "id": "85c85767-82b5-45e8-bc5e-fbb69a5738eb"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- c(2,  3,   5,    7)\n",
        "y <- c(1, 10, 100, 1000)\n",
        "x + y\n",
        "\n",
        "x * y"
      ],
      "id": "200533de-463d-4ce9-9148-8fbaec8eb7b8"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "sum(x * y)"
      ],
      "id": "f8238b9c-512f-458a-b9d2-bf4836b4ba2f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- c(TRUE, FALSE)\n",
        "y <- c(TRUE, TRUE)\n",
        "x & y"
      ],
      "id": "2621b552-d7d6-49d3-be58-169dcb49d485"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "u <- c(1, 2, 3)\n",
        "v <- c(1, 2, 3)\n",
        "w <- c(1, 2, 4)\n",
        "\n",
        "identical(u, v) # \u5168\u4f53\u306e\u6bd4\u8f03\n",
        "\n",
        "identical(u, w) # \u5168\u4f53\u306e\u6bd4\u8f03\n",
        "\n",
        "u == v          # \u8981\u7d20\u3054\u3068\u306e\u6bd4\u8f03\n",
        "\n",
        "u == w          # \u8981\u7d20\u3054\u3068\u306e\u6bd4\u8f03"
      ],
      "id": "6c290804-ee14-4f11-a615-ca41c710b222"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "sum(u == w)  # \u540c\u3058\u8981\u7d20\u306e\u6570\n",
        "\n",
        "mean(u == w) # \u540c\u3058\u8981\u7d20\u306e\u5272\u5408"
      ],
      "id": "5e80e6fc-d5a6-4d91-9913-6b01852e5b92"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- list(1, \"two\")"
      ],
      "id": "9b3d874c-595a-4d56-ba6d-b66fda85d307"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x[[2]]"
      ],
      "id": "1b0742fa-8907-4075-9e11-d65c41812397"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- list(\"apple\"  = \"\u308a\u3093\u3054\",\n",
        "          \"orange\" = \"\u307f\u304b\u3093\")"
      ],
      "id": "3e827f95-9612-445f-a7ec-be863af935bb"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x[[\"grape\"]] <- \"\u3076\u3069\u3046\""
      ],
      "id": "2fd47ff4-ec0f-4e93-8c0c-955304f1658c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x$apple\n",
        "# \u3042\u308b\u3044\u306f\n",
        "x$\"apple\"\n",
        "# \u3042\u308b\u3044\u306f\n",
        "x[[\"apple\"]]\n",
        "# \u3042\u308b\u3044\u306f\n",
        "tmp <- \"apple\"\n",
        "x[[tmp]]\n"
      ],
      "id": "7a33b198-2431-4a95-8b48-06b7331c11b2"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- c(\"foo\", \"bar\", \"baz\")\n",
        "y <- x\n",
        "y[2] <- \"BAR\" # y\u3092\u66f4\u65b0\u3059\u308b\uff0e\n",
        "y\n",
        "\n",
        "x             # x\u306f\u5909\u308f\u3089\u306a\u3044\uff0e"
      ],
      "id": "3f976917-41a0-4fa1-bb42-db7385ae2058"
    },
    {
      "id": "c13ab980",
      "cell_type": "markdown",
      "source": "## 3.4 \u30c7\u30fc\u30bf\u30d5\u30ec\u30fc\u30e0",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)"
      ],
      "id": "7e8cd35f-7c66-4117-8a82-c82c117ea5b4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df <- data.frame(\n",
        "  name    = c(\"A\", \"B\", \"C\", \"D\"),\n",
        "  english = c( 60,  90,  70,  90),\n",
        "  math    = c( 70,  80,  90, 100),\n",
        "  gender  = c(\"f\", \"m\", \"m\", \"f\"))"
      ],
      "id": "145ded96-34c9-4bb6-8fa3-f26e5b40eb02"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df <- tribble(\n",
        "  ~name, ~english, ~math, ~gender,\n",
        "  \"A\",         60,    70,     \"f\",\n",
        "  \"B\",         90,    80,     \"m\",\n",
        "  \"C\",         70,    90,     \"m\",\n",
        "  \"D\",         90,   100,     \"f\")"
      ],
      "id": "88a21e7d-877c-4916-b105-05b596cf836a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "head(my_df)\n",
        "# \u7d50\u679c\u306f\u5272\u611b"
      ],
      "id": "ae94eba1-d305-4f4d-9439-59093f9065c3"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "dim(my_df)  # \u884c\u6570\u3068\u5217\u6570\n",
        "\n",
        "nrow(my_df) # \u884c\u6570\n",
        "\n",
        "ncol(my_df) # \u5217\u6570"
      ],
      "id": "70bcd9e9-316d-4623-b200-f3f03a4165fe"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df2 <- expand.grid(\n",
        "  X = c(1, 2, 3),\n",
        "  Y = c(10, 100))\n",
        "my_df2"
      ],
      "id": "1dd1d45c-f928-4fdc-bae8-acbbae9f2b47"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "colnames(my_df2)"
      ],
      "id": "9fdc41d9-f213-4955-bda4-a7154ba48dba"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "colnames(my_df2) <- c(\"P\", \"Q\")\n",
        "my_df2\n",
        "# \u4ee5\u4e0b\u7701\u7565"
      ],
      "id": "99fa7c9c-4e23-41f6-a11c-85b4e433519b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "row.names(my_df)"
      ],
      "id": "f9c8deb0-7f00-4ddd-bbf0-317535bad16d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "row.names(my_df2) <-\n",
        "  c(\"a\", \"b\", \"c\", \"d\", \"e\", \"f\")\n",
        "my_df2\n",
        "# \u4ee5\u4e0b\u7701\u7565"
      ],
      "id": "fd920220-9aea-49b6-ae61-95f85e6768b7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df3 <- data.frame(\n",
        "  english =   c( 60,  90,  70,  90),\n",
        "  math    =   c( 70,  80,  90, 100),\n",
        "  gender  =   c(\"f\", \"m\", \"m\", \"f\"),\n",
        "  row.names = c(\"A\", \"B\", \"C\", \"D\"))\n",
        "my_df3"
      ],
      "id": "a561208f-de03-4407-a80c-7dd946d051ce"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- data.frame(\n",
        "  name    = \"E\",\n",
        "  english =  80,\n",
        "  math    =  80,\n",
        "  gender  = \"m\")\n",
        "my_df2 <- rbind(my_df, tmp)"
      ],
      "id": "232b397a-746d-4e4e-9d14-0115f9bc0f88"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df2 <- my_df %>%\n",
        "  mutate(id = c(1, 2, 3, 4))"
      ],
      "id": "b7089670-f55f-4281-b629-73a2239c5806"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df3 <- my_df               # \u30b3\u30d4\u30fc\n",
        "my_df3[\"id\"] <- c(1, 2, 3, 4) # \u66f4\u65b0\n",
        "my_df3 # \u7d50\u679c\u306e\u78ba\u8a8d\uff08\u5272\u611b\uff09"
      ],
      "id": "40507d12-8465-4371-9b31-bfb51a0a0268"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df[1, 2]"
      ],
      "id": "ecc18a04-b1b9-4834-a1da-8436d27ad9ce"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- my_df[, 2]\n",
        "# \u3042\u308b\u3044\u306f\n",
        "x <- my_df$english\n",
        "# \u3042\u308b\u3044\u306f\n",
        "x <- my_df$\"english\"\n",
        "# \u3042\u308b\u3044\u306f\n",
        "x <- my_df[[\"english\"]]\n",
        "# \u3042\u308b\u3044\u306f\n",
        "tmp <- \"english\"\n",
        "x <- my_df[[tmp]]\n",
        "\n",
        "x # \u7d50\u679c\u306e\u78ba\u8a8d\uff08\u5272\u611b\uff09"
      ],
      "id": "97ad08e6-19f8-48a0-bed8-d80b384f60a7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- my_df %>% select(name, math)"
      ],
      "id": "67bf29c5-4419-4f51-94ae-1bd88eee9b27"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- my_df[, c(1, 3)]"
      ],
      "id": "065c7877-3bc8-4cc8-b9e6-3404e568dc92"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- my_df %>%\n",
        "  select(-c(english, gender))\n",
        "# \u3042\u308b\u3044\u306f\n",
        "x <- my_df[, -c(2, 4)]"
      ],
      "id": "9b7adff3-4ec7-494d-8332-ccffcd31f9fc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- my_df[c(1, 3), ]"
      ],
      "id": "4bc0997b-f831-4360-a52f-a8adaa28ae7e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- my_df[-c(2, 4), ]"
      ],
      "id": "821983cb-b4c1-4eb0-acf7-6e12b7a95780"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- my_df[my_df$gender == \"m\", ]\n",
        "# \u3042\u308b\u3044\u306f\n",
        "x <- my_df %>% filter(gender == \"m\")"
      ],
      "id": "69a56524-6c01-46b8-8fb5-af6faacb983c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- my_df[my_df$english > 80 & my_df$gender == \"m\", ]\n",
        "# \u3042\u308b\u3044\u306f\n",
        "x <- my_df %>% filter(english > 80 & gender == \"m\")"
      ],
      "id": "213f95ac-c2f9-4d90-a386-db577330fd61"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- my_df[my_df$english == max(my_df$english), ]\n",
        "# \u3042\u308b\u3044\u306f\n",
        "x <- my_df %>% filter(english == max(my_df$english))"
      ],
      "id": "e52752f4-42e2-48c7-8386-07549bbc78dc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df2 <- my_df # \u30b3\u30d4\u30fc\n",
        "my_df2[my_df$gender == \"m\", ]$gender <- \"M\""
      ],
      "id": "0eaa1c31-8d40-44b7-9e76-c2d09ac7c9ca"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df2"
      ],
      "id": "f2ed4ccf-ad7a-43f9-8cc4-2f48f777ebe3"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- my_df %>% arrange(english)"
      ],
      "id": "e2aaeb60-2c0d-4489-94ea-d7ed201e54fd"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- my_df %>% arrange(-english)"
      ],
      "id": "74e34e43-6ff1-4f6f-94c7-bb2f115416da"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- c(2, 3, 5, 7, 11, 13, 17, 19, 23,\n",
        "       29, 31, 37)\n",
        "A <- matrix(\n",
        "  data = x,     # 1\u6b21\u5143\u30c7\u30fc\u30bf\n",
        "  nrow = 3,     # \u884c\u6570\n",
        "  byrow = TRUE) # \u884c\u3054\u3068\u306e\u751f\u6210\n",
        "A"
      ],
      "id": "ee5720e8-742a-4efe-904d-eb58255d6fc6"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "A <- my_df[, c(2, 3)] %>% as.matrix\n",
        "A"
      ],
      "id": "f53ac63a-a33b-4950-aeeb-cae1557281dc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "as.data.frame(A)"
      ],
      "id": "be848dfa-54f9-436e-9be7-1b8f944800d7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "t(A)"
      ],
      "id": "96a2cb69-fe60-427d-a1fb-0ba179434413"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "t(A) %*% A"
      ],
      "id": "4aa81944-0c82-4d8a-9cbe-fc8975b5ed17"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_wider <- data.frame(\n",
        "  day = c(25, 26, 27),\n",
        "  min = c(20, 21, 15),\n",
        "  max = c(24, 27, 21))"
      ],
      "id": "12440534-6cef-4ae1-b70c-5e119f77be49"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_longer <- my_wider %>%\n",
        "  pivot_longer(-day)\n",
        "my_longer"
      ],
      "id": "8c0bb525-4bcd-4b26-af5c-d17724b71b82"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_longer %>% pivot_wider()"
      ],
      "id": "ebb17dd2-036b-4943-98cd-94d12f24f4ef"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_longer %>%\n",
        "  ggplot(aes(x = day, y = value,\n",
        "             color = name)) +\n",
        "  geom_point() +\n",
        "  geom_line() +\n",
        "  ylab(\"temperature\") + # y\u8ef8\u30e9\u30d9\u30eb\n",
        "  scale_x_continuous(\n",
        "    breaks = my_longer$day) # x\u8ef8\u76ee\u76db\u308a"
      ],
      "id": "1ab984c2-1ea5-4e7c-aa64-7f36f55e335a"
    },
    {
      "id": "17f67904",
      "cell_type": "markdown",
      "source": "## 3.5 1\u6b21\u5143\u30c7\u30fc\u30bf\u306e\uff08\u975e\uff09\u985e\u4f3c\u5ea6",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "A <- c(3,   4,  5)\n",
        "B <- c(3,   4, 29)\n",
        "C <- c(9, -18,  8)\n",
        "AB <- B - A\n",
        "AC <- C - A\n",
        "\n",
        "sum(AB^2)^0.5\n",
        "\n",
        "sum(AC^2)^0.5"
      ],
      "id": "0142a366-7d6f-4e5c-a9f7-b02bd263424b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "sum(abs(AB))\n",
        "\n",
        "sum(abs(AC))"
      ],
      "id": "5169cd8b-a2ee-4767-9093-57e142bf2195"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "sum(A * B) /\n",
        "  sum(A * A)^0.5 / sum(B * B)^0.5\n",
        "\n",
        "sum(A * C) /\n",
        "  sum(A * A)^0.5 / sum(C * C)^0.5"
      ],
      "id": "7f87299b-ee2f-452f-96c4-0e25e4037059"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "cor(A, B)\n",
        "\n",
        "cor(A, C)"
      ],
      "id": "d6b67401-4e55-41c3-80c4-e025a4d23908"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "\n",
        "\n",
        "\n",
        "my_df <- data.frame(\n",
        "  x = c(3,  3,   9),\n",
        "  y = c(4,  4, -18),\n",
        "  z = c(5, 29,   8),\n",
        "  row.names = c(\"A\", \"B\", \"C\"))\n",
        "\n",
        "# \u30e6\u30fc\u30af\u30ea\u30c3\u30c9\u8ddd\u96e2\n",
        "my_df %>% proxy::dist(\"Euclidean\")\n",
        "\n",
        "\n",
        "# \u30de\u30f3\u30cf\u30c3\u30bf\u30f3\u8ddd\u96e2\n",
        "my_df %>% proxy::dist(\"Manhattan\")\n",
        "\n",
        "\n",
        "# \u30b3\u30b5\u30a4\u30f3\u985e\u4f3c\u5ea6\n",
        "my_df %>% proxy::simil(\"cosine\")\n",
        "\n",
        "\n",
        "# \u76f8\u95a2\u4fc2\u6570\n",
        "my_df %>% proxy::simil(\"correlation\")"
      ],
      "id": "ef192a43-e703-4f6a-b880-866a0f79c831"
    },
    {
      "id": "f469f739",
      "cell_type": "markdown",
      "source": "## 3.6 R\u306e\u30d1\u30c3\u30b1\u30fc\u30b8\uff0cPython\u306e\u30e2\u30b8\u30e5\u30fc\u30eb",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)"
      ],
      "id": "0e3291fd-3656-4f79-adbd-093f33d8daa3"
    },
    {
      "id": "77867623",
      "cell_type": "markdown",
      "source": "## 3.7 \u53cd\u5fa9\u51e6\u7406",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)"
      ],
      "id": "187fc384-7e27-44d5-bdb8-91a45a541e5d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "f1 <- function(x) {\n",
        "  tmp <- runif(x)\n",
        "  mean(tmp)\n",
        "}\n",
        "\n",
        "f1(10)           # \u52d5\u4f5c\u78ba\u8a8d"
      ],
      "id": "b351e561-2fd6-4791-b65f-313757f57afc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "replicate(n = 3, expr = f1(10))"
      ],
      "id": "7706b322-7141-496e-b5b9-399ec82abcbe"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "rep(x = f1(10), times = 3)"
      ],
      "id": "14038f15-fd77-421e-933c-b46e92de435e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "v <- c(5, 10, 100)\n",
        "v %>% map_dbl(f1)"
      ],
      "id": "1d219fd3-629d-423b-a8ba-41a2bb304da1"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "rep(x = 10, times = 3) %>% map_dbl(f1)\n",
        "# \u7d50\u679c\u306f\u5272\u611b"
      ],
      "id": "49b60f27-18f5-4792-b780-957b616d87a5"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "f2 <- function(n) {\n",
        "  tmp <- runif(n)\n",
        "  list(x = n,\n",
        "       p = mean(tmp),\n",
        "       q = sd(tmp))\n",
        "}\n",
        "\n",
        "f2(10) # \u52d5\u4f5c\u78ba\u8a8d"
      ],
      "id": "36f4812e-5898-454f-a684-90d7a08e0ccd"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "v <- c(5, 10, 100)\n",
        "v %>% map_dfr(f2)"
      ],
      "id": "2cd415e7-8410-4c6d-a7cc-38d2323ddb47"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "f3 <- function(x, y) {\n",
        "  tmp <- runif(x, min = 1,\n",
        "                  max = y + 1) %>%\n",
        "    as.integer\n",
        "  list(x = x,\n",
        "       y = y,\n",
        "       p = mean(tmp),\n",
        "       q = sd(tmp))\n",
        "}\n",
        "\n",
        "f3(x = 10, y = 6) # \u52d5\u4f5c\u78ba\u8a8d"
      ],
      "id": "f544fb39-80bf-4418-88b4-f6a3cb9d925c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df <- data.frame(\n",
        "  x = c(5, 10, 100,  5, 10, 100),\n",
        "  y = c(6,  6,   6, 12, 12,  12))\n",
        "\n",
        "my_df %>% pmap_dfr(f3)"
      ],
      "id": "9a6d19b1-b7e6-4cd3-86e2-da2a0ee9cb57"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(furrr)\n",
        "plan(multisession) # \u6e96\u5099\n",
        "\n",
        "v <- c(5, 10, 100)\n",
        "v %>% future_map_dbl(f1, .options =\n",
        "  furrr_options(seed = TRUE))\n",
        "# \u7d50\u679c\u306f\u5272\u611b"
      ],
      "id": "21dafbe4-bfaa-4a4b-8c68-7e805149c44f"
    },
    {
      "id": "1b9c303f",
      "cell_type": "markdown",
      "source": "## 3.8 \u305d\u306e\u4ed6",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- 123\n",
        "typeof(x)"
      ],
      "id": "cdfb47a0-3cd7-4182-9953-bdeec65d8020"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "?log\n",
        "# \u3042\u308b\u3044\u306f\n",
        "help(log)"
      ],
      "id": "43e0f294-d752-4c53-9597-86fdac9a5ce4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "v <- c(1, NA, 3)\n",
        "v"
      ],
      "id": "afc52099-a152-42eb-86ef-4a59fb5c3498"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "is.na(v[2])\n",
        "\n",
        "v[2] == NA # \u8aa4\u308a"
      ],
      "id": "d8a529f6-1f4f-4ea3-a2f3-ea6f2f560d9b"
    },
    {
      "id": "eeaae488",
      "cell_type": "markdown",
      "source": "# 4 \u7d71\u8a08\u5165\u9580",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Google Colaboratory\u306e\u74b0\u5883\u8a2d\u5b9a\n",
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  options(Ncpus = parallel::detectCores())\n",
        "  installed_packages <- rownames(installed.packages())\n",
        "  packages_to_install <- c(\"exactci\", \"ggmosaic\", \"pastecs\", \"psych\", \"vcd\")\n",
        "  install.packages(setdiff(packages_to_install, installed_packages))\n",
        "}"
      ],
      "id": "1bc6ca8d-2107-4930-9a62-4bb154a9dd81"
    },
    {
      "id": "f1585303",
      "cell_type": "markdown",
      "source": "## 4.1 \u8a18\u8ff0\u7d71\u8a08",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- c(165, 170, 175, 180, 185)\n",
        "mean(x) # \u5e73\u5747"
      ],
      "id": "9ba92b45-d8f7-4cf4-afeb-4b1b4e86f4b1"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "n <- length(x) # \u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\n",
        "sum(x) / n"
      ],
      "id": "732b0eb9-536e-4c5d-be79-65105df4fd68"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y <- c(173, 174, 175, 176, 177)\n",
        "mean(y)"
      ],
      "id": "b5cf31d1-661f-420d-81e5-720375b2251e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "var(x) # x\u306e\u5206\u6563\n",
        "\n",
        "var(y) # y\u306e\u5206\u6563"
      ],
      "id": "9da21e67-01da-4112-8a1b-1b0c31a41e8b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "sum((x - mean(x))^2) / (n - 1)"
      ],
      "id": "90a0bb0b-403d-42a8-8720-186093882bad"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "sd(x) # x\u306e\u6a19\u6e96\u504f\u5dee\n",
        "\n",
        "sd(y) # y\u306e\u6a19\u6e96\u504f\u5dee"
      ],
      "id": "a9d3183b-f46e-4994-bc0f-2d64e63624d5"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "var(x)**0.5 # x\u306e\u6a19\u6e96\u504f\u5dee"
      ],
      "id": "6d37239b-abdd-4a30-ae15-af01e31cb854"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "psych::describe(x)\n",
        "\n",
        "# \u3042\u308b\u3044\u306f\n",
        "\n",
        "pastecs::stat.desc(x)"
      ],
      "id": "b0ac85f2-5cab-4a94-aed2-9ff986a36a3e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "quantile(x)"
      ],
      "id": "15e081da-b30c-44ad-bd77-3a17f59fcd4a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- c(165, 170, 175, 180, 185)\n",
        "\n",
        "var(x)                # \u4e0d\u504f\u5206\u6563\n",
        "\n",
        "mean((x - mean(x))^2) # \u6a19\u672c\u5206\u6563\n",
        "# \u3042\u308b\u3044\u306f\n",
        "n <- length(x)\n",
        "var(x) * (n - 1) / n  # \u6a19\u672c\u5206\u6563"
      ],
      "id": "a6389cf3-e037-4253-9581-46f7dd9faaba"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "sd(x)                     # \u221a\u4e0d\u504f\u5206\u6563\n",
        "\n",
        "mean((x - mean(x))^2)^0.5 # \u221a\u6a19\u672c\u5206\u6563\n",
        "# \u3042\u308b\u3044\u306f\n",
        "sd(x) * sqrt((n - 1) / n) # \u221a\u6a19\u672c\u5206\u6563"
      ],
      "id": "9a2ab9cb-4993-4e23-8fd5-fc64337cc8e4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "sd(x) / length(x)**0.5"
      ],
      "id": "51d82a42-9197-4515-b866-1b93a68f74d4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "\n",
        "my_df <- data.frame(\n",
        "  name    = c(\"A\", \"B\", \"C\", \"D\"),\n",
        "  english = c( 60,  90,  70,  90),\n",
        "  math    = c( 70,  80,  90, 100),\n",
        "  gender  = c(\"f\", \"m\", \"m\", \"f\"))"
      ],
      "id": "84f2b867-da79-4d8b-a039-b0b88f396b27"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "var(my_df$english)"
      ],
      "id": "3bab0b68-13e6-49fe-9777-96f103271bf4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u7d50\u679c\u306f\u30d9\u30af\u30bf\n",
        "my_df[, c(2, 3)] %>% sapply(var)\n",
        "\n",
        "# \u7d50\u679c\u306f\u30ea\u30b9\u30c8\n",
        "my_df[, c(2, 3)] %>% lapply(var)\n",
        "\n",
        "# \u7d50\u679c\u306f\u30c7\u30fc\u30bf\u30d5\u30ec\u30fc\u30e0\n",
        "my_df[, c(2, 3)] %>% # 2, 3\u5217\u76ee\n",
        "  summarize(across(  # \u306e\n",
        "    everything(),    # \u5168\u3066\u306e\n",
        "    var))            # \u4e0d\u504f\u5206\u6563\n",
        "# \u3042\u308b\u3044\u306f\n",
        "my_df %>%              # \u30c7\u30fc\u30bf\u30d5\u30ec\u30fc\u30e0\n",
        "  summarize(across(    # \u306e\n",
        "    where(is.numeric), # \u6570\u5024\u306e\u5217\u306e\n",
        "    var))              # \u4e0d\u504f\u5206\u6563\n",
        "# \u3042\u308b\u3044\u306f\n",
        "my_df %>%              # \u30c7\u30fc\u30bf\u30d5\u30ec\u30fc\u30e0\n",
        "  summarize(across(    # \u306e\n",
        "    where(is.numeric), # \u6570\u5024\u306e\u5217\u306e\n",
        "    function(x) { var(x) })) # \u4e0d\u504f\u5206\u6563\n"
      ],
      "id": "42f467d7-19d8-4f44-88ea-52d4e7175642"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "psych::describe(my_df)\n",
        "\n",
        "# \u3042\u308b\u3044\u306f\n",
        "\n",
        "pastecs::stat.desc(my_df)\n",
        "# \u4ee5\u4e0b\u7701\u7565"
      ],
      "id": "241a942b-f455-4c37-8878-0670707fb766"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "table(my_df$gender)\n"
      ],
      "id": "9bba615e-f295-4e26-ab1b-f98fad87a7c7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df2 <- data.frame(\n",
        "  gender = my_df$gender,\n",
        "  excel = my_df$math >= 80)\n",
        "table(my_df2)\n"
      ],
      "id": "18a04dfd-c340-4835-8641-d197a717dcc8"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df %>% group_by(gender) %>%\n",
        "  summarize(across(\n",
        "    where(is.numeric), mean),\n",
        "    .groups = \"drop\") # \u30b0\u30eb\u30fc\u30d7\u5316\u89e3\u9664\n"
      ],
      "id": "e577ad89-e6a5-45e5-bf84-27f467597551"
    },
    {
      "id": "e8b863c8",
      "cell_type": "markdown",
      "source": "## 4.2 \u30c7\u30fc\u30bf\u306e\u53ef\u8996\u5316",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "head(iris)"
      ],
      "id": "883dfe20-69ee-4c1a-bd1a-0d354b30ecec"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "hist(iris$Sepal.Length)"
      ],
      "id": "0b749b0f-e2f1-4d98-807d-a011841d5445"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- c(10, 20, 30)\n",
        "hist(x, breaks = 2) # \u968e\u7d1a\u6570\u306f2"
      ],
      "id": "3829e10d-87b7-41b0-bf40-25de1ee5bda1"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- iris$Sepal.Length\n",
        "tmp <- seq(min(x), max(x),\n",
        "           length.out = 10)\n",
        "hist(x, breaks = tmp, right = FALSE)"
      ],
      "id": "5e5ded28-ef53-4c55-92e9-9be395efe23c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "plot(iris$Sepal.Length,\n",
        "     iris$Sepal.Width)"
      ],
      "id": "be774a5b-2e7b-43fd-86a7-29b547706190"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "boxplot(iris[, -5])"
      ],
      "id": "4341c44e-ada8-49bb-9c1d-90d2e6d67573"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "my_df <- psych::describe(iris[, -5])\n",
        "my_df %>% select(mean, sd, se)"
      ],
      "id": "bb895ee1-ba11-4fdc-94e4-d0ab19cd0f2d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- rownames(my_df)\n",
        "my_df %>% ggplot(aes(x = factor(tmp, levels = tmp), y = mean)) +\n",
        "  geom_col() +\n",
        "  geom_errorbar(aes(ymin = mean - se, ymax = mean + se)) +\n",
        "  xlab(NULL)"
      ],
      "id": "b375ed6e-9ab7-48e0-ab5b-c8380bd76b66"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_group <- iris %>% group_by(Species)       # \u54c1\u7a2e\u3054\u3068\u306b\uff0c\n",
        "\n",
        "my_df <- my_group %>%                        # \u5404\u5909\u6570\u306e\uff0c\u5e73\u5747\u3068\n",
        "  summarize(across(everything(), mean)) %>%\n",
        "  pivot_longer(-Species)\n",
        "\n",
        "tmp <- my_group %>%                          # \u6a19\u6e96\u8aa4\u5dee\u3092\u6c42\u3081\u308b\uff0e\n",
        "  summarize(across(everything(), ~ sd(.) / length(.)**0.5)) %>%\n",
        "  pivot_longer(-Species)\n",
        "\n",
        "my_df$se <- tmp$value\n",
        "head(my_df)"
      ],
      "id": "3f41f21d-b059-4532-8069-a3470f8f2126"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df %>%\n",
        "  ggplot(aes(x = Species, y = value, fill = name)) +\n",
        "  geom_col(position = \"dodge\") +\n",
        "  geom_errorbar(aes(ymin = value - se, ymax = value + se), position = \"dodge\")"
      ],
      "id": "97b382ee-0459-4e30-9c91-2ff67d71a01c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u5404\u5909\u6570\u306e\u5e73\u5747\n",
        "iris %>% pivot_longer(-Species) %>%\n",
        "  ggplot(aes(x = name, y = value)) +\n",
        "  geom_bar(stat = \"summary\", fun = mean) +\n",
        "  stat_summary(geom = \"errorbar\", fun.data = mean_se) +\n",
        "  xlab(NULL)\n",
        "\n",
        "# \u5404\u5909\u6570\u306e\u5e73\u5747\uff08\u54c1\u7a2e\u3054\u3068\uff09\n",
        "iris %>% pivot_longer(-Species) %>%\n",
        "  ggplot(aes(x = Species, y = value, fill = name)) +\n",
        "  geom_bar(stat = \"summary\", fun = mean, position = \"dodge\") +\n",
        "  stat_summary(geom = \"errorbar\", fun.data = mean_se, position = \"dodge\")"
      ],
      "id": "8d110e54-05ce-4778-993e-34b1d7faa33e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df <- data.frame(\n",
        "  Species = iris$Species,\n",
        "  w_Sepal = iris$Sepal.Width > 3)\n",
        "table(my_df) # \u5206\u5272\u8868\n",
        "\n",
        "mosaicplot(\n",
        "  formula = ~ Species + w_Sepal,\n",
        "  data = my_df)"
      ],
      "id": "99b444a3-e8d1-4682-a85a-757b4b2cc440"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(vcd)\n",
        "vcd::mosaic(formula = ~w_Sepal + Species, data = my_df,\n",
        "            labeling = labeling_values)"
      ],
      "id": "8212beb2-11fe-4917-871c-546effbd7af1"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "curve(x^3 - x, -2, 2)"
      ],
      "id": "995e45f7-e9bc-4f81-84ed-a2d9dae2786f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- iris$Sepal.Length\n",
        "tmp <- seq(min(x), max(x),\n",
        "           length.out = 10)\n",
        "iris %>%\n",
        "  ggplot(aes(x = Sepal.Length)) +\n",
        "  geom_histogram(breaks = tmp,\n",
        "                 closed = \"left\")"
      ],
      "id": "01a7186d-8453-45ca-810c-d88245377cdc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "iris %>%\n",
        "  ggplot(aes(x = Sepal.Length,\n",
        "             y = Sepal.Width)) +\n",
        "  geom_point()"
      ],
      "id": "bb8dd24d-5c1f-4fa7-9cbc-dbd91e6de2cc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "iris %>%\n",
        "  pivot_longer(-Species) %>%\n",
        "  ggplot(aes(\n",
        "    x = factor(name,\n",
        "               levels = names(iris)),\n",
        "    y = value)) +\n",
        "  geom_boxplot() +\n",
        "  xlab(NULL)"
      ],
      "id": "830d90c8-62a3-4821-8e78-a1c5a7b697ae"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(ggmosaic)\n",
        "my_df <- data.frame(\n",
        "  Species = iris$Species,\n",
        "  w_Sepal = iris$Sepal.Width > 3)\n",
        "my_df %>%\n",
        "  ggplot() +\n",
        "  geom_mosaic(\n",
        "    aes(x = product(w_Sepal, Species)))"
      ],
      "id": "55da998c-7352-440e-9d88-a84e83e58104"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "f <- function(x) { x^3 - x }\n",
        "data.frame(x = c(-2, 2)) %>%\n",
        "  ggplot(aes(x = x)) +\n",
        "  stat_function(fun = f)"
      ],
      "id": "a35d81f7-58b8-42cd-8575-4d678938bf97"
    },
    {
      "id": "0bb9f790",
      "cell_type": "markdown",
      "source": "## 4.3 \u4e71\u6570",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- sample(x = 1:6,        # \u7bc4\u56f2\n",
        "            size = 10000,   # \u4e71\u6570\u306e\u6570\n",
        "            replace = TRUE) # \u91cd\u8907\u3042\u308a\n",
        "hist(x, breaks = 0:6) # \u30d2\u30b9\u30c8\u30b0\u30e9\u30e0"
      ],
      "id": "f7b618fc-71ca-4042-8bf4-c91a32f8a5ef"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- runif(min = 0,  # \u6700\u5c0f\n",
        "           max = 1,  # \u6700\u5927\n",
        "           n = 1000) # \u4e71\u6570\u306e\u6570\n",
        "hist(x)"
      ],
      "id": "79572b8c-8362-4aa1-befb-132ee1a5986c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x <- as.integer(      # \u6574\u6570\u306b\u5909\u63db\n",
        "  runif(min = 1,      # \u6700\u5c0f\n",
        "        max = 7,      # \u6700\u5927 + 1\n",
        "        n = 1000))    # \u4e71\u6570\u306e\u6570\n",
        "hist(x, breaks = 0:6) # \u7d50\u679c\u306f\u5272\u611b"
      ],
      "id": "e91fa1ba-388f-4165-a42b-98238441c479"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "n <- 100\n",
        "p <- 0.5\n",
        "r <- 10000\n",
        "x <- rbinom(size = n, # \u8a66\u884c\u56de\u6570\n",
        "            prob = p, # \u78ba\u7387\n",
        "            n = r)    # \u4e71\u6570\u306e\u6570\n",
        "hist(x, breaks = max(x) - min(x))"
      ],
      "id": "dab47d8e-a90f-4449-ae15-baf598d195cc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "r <- 10000\n",
        "x <- rnorm(mean = 50, # \u5e73\u5747\n",
        "           sd = 5,    # \u6a19\u6e96\u504f\u5dee\n",
        "           n = r)     # \u4e71\u6570\u306e\u6570\n",
        "hist(x, breaks = 40)"
      ],
      "id": "c6141dad-cc4b-4a27-b267-06ff5fb51d47"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "\n",
        "f <- function(k) {\n",
        "  n <- 10000\n",
        "  tmp <- replicate(n = n, expr = g(rnorm(n =  k, sd = 3)))\n",
        "  list(k = k,\n",
        "       mean = mean(tmp),       # \u5e73\u5747\n",
        "       se = sd(tmp) / sqrt(n)) # \u6a19\u6e96\u8aa4\u5dee\n",
        "}"
      ],
      "id": "f0037339-566c-466c-a61d-578887fb4e19"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "g <- var\n",
        "c(10, 20, 30) %>% map_dfr(f)"
      ],
      "id": "54316139-062a-4547-b42b-feb443976ed0"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "g <- sd\n",
        "c(5, 10, 15, 20) %>% map_dfr(f)"
      ],
      "id": "3c314402-bed9-430f-b3c6-ff6c3c1040c7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "g <- function(x) {\n",
        "  n <- length(x)\n",
        "  sd(x) *\n",
        "    sqrt((n - 1) / 2) *\n",
        "    gamma((n - 1) / 2) /\n",
        "    gamma(n / 2)\n",
        "}\n",
        "c(10, 20, 30) %>% map_dfr(f)"
      ],
      "id": "5f58d7ca-0378-45ce-876c-51b399395d41"
    },
    {
      "id": "aabb17fc",
      "cell_type": "markdown",
      "source": "## 4.4 \u7d71\u8a08\u7684\u63a8\u6e2c",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(exactci)\n",
        "library(tidyverse)\n",
        "\n",
        "a <- 0.05                              # \u6709\u610f\u6c34\u6e96\n",
        "binom.exact(x = 2,                     # \u5f53\u305f\u3063\u305f\u56de\u6570\n",
        "            n = 15,                    # \u304f\u3058\u3092\u5f15\u3044\u305f\u56de\u6570\n",
        "            p = 4 / 10,                # \u5f53\u305f\u308b\u78ba\u7387\uff08\u4eee\u8aac\uff09\n",
        "            plot = TRUE,               # p\u5024\u306e\u63cf\u753b\uff08\u7d50\u679c\u306f\u6b21\u9805\u306b\u63b2\u8f09\uff09\n",
        "            conf.level = 1 - a,        # \u4fe1\u983c\u4fc2\u6570\uff08\u30c7\u30d5\u30a9\u30eb\u30c8\uff09\n",
        "            tsmethod = \"minlike\",      # p\u5024\u306e\u5b9a\u7fa9\n",
        "            alternative = \"two.sided\") # \u4e21\u5074\u691c\u5b9a\uff08\u30c7\u30d5\u30a9\u30eb\u30c8\uff09\n",
        "                                       # \u5de6\u7247\u5074\u691c\u5b9a\u306a\u3089'less'\n",
        "                                       # \u53f3\u7247\u5074\u691c\u5b9a\u306a\u3089'greater'\n"
      ],
      "id": "43294893-42c2-41d3-a13f-17633f58ef19"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "t <- 4 / 10               # \u5f53\u305f\u308b\u78ba\u7387\n",
        "n <- 15                   # \u304f\u3058\u3092\u5f15\u3044\u305f\u56de\u6570\n",
        "x <- 0:n                  # \u5f53\u305f\u3063\u305f\u56de\u6570\n",
        "my_pr  <- dbinom(x, n, t) # x\u56de\u5f53\u305f\u308b\u78ba\u7387\n",
        "my_pr2 <- dbinom(2, n, t) # 2\u56de\u5f53\u305f\u308b\u78ba\u7387\n",
        "\n",
        "my_data <- data.frame(x = x,\n",
        "                      probability = my_pr,\n",
        "                      color = my_pr <= my_pr2) # \u5f53\u305f\u308b\u78ba\u7387\u304c\uff0c2\u56de\u5f53\u305f\u308b\u78ba\u7387\u4ee5\u4e0b\n",
        "\n",
        "my_data %>% ggplot(aes(x = x, y = probability, color = color)) +\n",
        "  geom_point(size = 3) +\n",
        "  geom_linerange(aes(ymin = 0, ymax = probability), ) + # \u5782\u76f4\u7dda\n",
        "  geom_hline(yintercept = my_pr2) +                     # \u6c34\u5e73\u7dda\n",
        "  theme(legend.position = \"none\")                       # \u51e1\u4f8b\u3092\u8868\u793a\u3057\u306a\u3044\uff0e"
      ],
      "id": "0dee0fcf-5b29-406b-9813-8e3903e0fd4a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u524d\u9805\u306e\u7d50\u679c\uff08\u518d\u63b2\uff09"
      ],
      "id": "0e5de86e-58e1-48fd-aa6c-023aef7bcf00"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u524d\u9805\u5192\u982d\u306e\u30b3\u30fc\u30c9"
      ],
      "id": "aa85b71d-a4fe-439c-bb6a-59fbd1d12959"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "X <- c(32.1, 26.2, 27.5, 31.8, 32.1, 31.2, 30.1, 32.4, 32.3, 29.9,\n",
        "       29.6, 26.6, 31.2, 30.9, 29.3)\n",
        "Y <- c(35.4, 34.6, 31.1, 32.4, 33.3, 34.7, 35.3, 34.3, 32.1, 28.3,\n",
        "       33.3, 30.5, 32.6, 33.3, 32.2)\n",
        "\n",
        "t.test(x = X, y = Y,\n",
        "       conf.level = 0.95,         # \u4fe1\u983c\u4fc2\u6570\uff08\u30c7\u30d5\u30a9\u30eb\u30c8\uff09\n",
        "       paired = TRUE,             # \u5bfe\u6a19\u672c\u3067\u3042\u308b\uff0e\n",
        "       alternative = \"two.sided\") # \u4e21\u5074\u691c\u5b9a\uff08\u30c7\u30d5\u30a9\u30eb\u30c8\uff09\n",
        "                                  # \u5de6\u7247\u5074\u691c\u5b9a\u306a\u3089'less'\n",
        "                                  # \u53f3\u7247\u5074\u691c\u5b9a\u306a\u3089'greater'\n"
      ],
      "id": "f6ffacaf-2585-4b13-ae67-2cc30bf771f9"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "t.test(x = X, y = Y,\n",
        "       paired = FALSE,   # \u5bfe\u6a19\u672c\u3067\u306f\u306a\u3044\uff08\u30c7\u30d5\u30a9\u30eb\u30c8\uff09\uff0e\n",
        "       var.equal = TRUE, # \u7b49\u5206\u6563\u3092\u4eee\u5b9a\u3059\u308b\uff0e\u4eee\u5b9a\u3057\u306a\u3044\u306a\u3089FALSE\uff08\u30c7\u30d5\u30a9\u30eb\u30c8\uff09\uff0e\n",
        "       alternative = \"two.sided\",\n",
        "       conf.level = 0.95)\n"
      ],
      "id": "bee03a85-3f86-430a-9a5c-1b923d8e9150"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/smoker.csv\")\n",
        "my_data <- read_csv(my_url)"
      ],
      "id": "430bcdb1-7408-4cea-81fd-cea615777258"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "head(my_data)"
      ],
      "id": "ade4d3c1-f89e-4c78-b539-e30010948e13"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_table <- table(my_data)\n",
        "my_table"
      ],
      "id": "c27dff41-c70e-417f-bdbb-c497158a63f2"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "chisq.test(my_table, correct = FALSE)\n"
      ],
      "id": "5ebc2db4-c1ef-4ca0-8a7e-5ee04b8fe38d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "X <- rep(0:1, c(13, 2)) # \u624b\u98061\n",
        "X\n",
        "\n",
        "tmp <- sample(X, size = length(X), replace = TRUE) # \u624b\u98062\n",
        "tmp\n",
        "\n",
        "sum(tmp) # \u624b\u98063\n",
        "\n",
        "n <- 10^5\n",
        "result <- replicate(n, sum(sample(X, size = length(X), replace = TRUE))) # \u624b\u98064"
      ],
      "id": "1368aa64-81e2-4283-a3b3-c3181a4983c8"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "hist(x = result, breaks = 0:15,\n",
        "     right = FALSE)"
      ],
      "id": "b9792825-bbc7-4bb4-8163-501bed8015f4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "quantile(result, c(0.025, 0.975))"
      ],
      "id": "7c6d2f8b-0d69-449c-a12e-109d12d09d69"
    },
    {
      "id": "602a7989",
      "cell_type": "markdown",
      "source": "# 5 \u524d\u51e6\u7406",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Google Colaboratory\u306e\u74b0\u5883\u8a2d\u5b9a\n",
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  options(Ncpus = parallel::detectCores())\n",
        "  installed_packages <- rownames(installed.packages())\n",
        "  packages_to_install <- c(\"caret\")\n",
        "  install.packages(setdiff(packages_to_install, installed_packages))\n",
        "}"
      ],
      "id": "dd094285-faf6-459a-8a17-10a931921bf2"
    },
    {
      "id": "61066208",
      "cell_type": "markdown",
      "source": "## 5.1 \u30c7\u30fc\u30bf\u306e\u8aad\u307f\u8fbc\u307f",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "system(str_c(\"wget https://raw.githubusercontent.com/taroyabuki\",\n",
        "             \"/fromzero/master/data/exam.csv\"))"
      ],
      "id": "d4743cf4-dda7-4c5e-a268-5883b05d2046"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df <- read_csv(\"exam.csv\")\n",
        "# \u3042\u308b\u3044\u306f\n",
        "my_df <- read.csv(\"exam.csv\",\n",
        "  stringsAsFactors = FALSE)\n",
        "\n",
        "my_df"
      ],
      "id": "cdd5c8fe-8947-4d05-bb9d-bdaca04d76ed"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/exam.csv\")\n",
        "my_df <- read_csv(my_url)\n",
        "# \u3042\u308b\u3044\u306f\n",
        "my_df <- read.csv(my_url, stringsAsFactors = FALSE)"
      ],
      "id": "9baa8030-1a7a-4f57-9b4b-7ad6f8e5be31"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df2 <- read.csv(\n",
        "  file = \"exam.csv\",\n",
        "  stringsAsFactors = FALSE,\n",
        "  row.names = 1)\n",
        "my_df2"
      ],
      "id": "cc288ce8-37d7-4b14-b4b1-af470ba27a81"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df %>% write_csv(\"exam2.csv\")\n",
        "# \u3042\u308b\u3044\u306f\n",
        "my_df %>% write.csv(\n",
        "  file = \"exam2.csv\",\n",
        "  row.names = FALSE)"
      ],
      "id": "40af9853-22d9-4149-8e13-ba99e0b7ca0c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df2 %>% write.csv(\"exam3.csv\")"
      ],
      "id": "515e01f5-c0b0-4815-bf01-6314fea34188"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df <- read_csv(file = \"exam.csv\",\n",
        "  locale = locale(encoding = \"UTF-8\"))\n",
        "# \u3042\u308b\u3044\u306f\n",
        "my_df <- read.csv(file = \"exam.csv\",\n",
        "  stringsAsFactors = FALSE,\n",
        "  fileEncoding = \"UTF-8\")"
      ],
      "id": "e5257494-f937-4c51-9220-2a911a75554a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df %>% write_csv(\"exam2.csv\")\n",
        "# \u3042\u308b\u3044\u306f\n",
        "my_df %>% write.csv(file = \"exam2.csv\", row.names = FALSE,\n",
        "                    fileEncoding = \"UTF-8\")"
      ],
      "id": "79b5a5c7-22c6-49e5-96bb-1308fd5f5882"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_url <- \"https://taroyabuki.github.io/fromzero/exam.html\"\n",
        "my_tables <- xml2::read_html(my_url) %>% rvest::html_table()"
      ],
      "id": "e5126f61-f7bc-45e5-a245-e079d840765c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_tables"
      ],
      "id": "1189b21b-c48e-4eb9-9018-ecec489729aa"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- my_tables[[1]]\n",
        "tmp"
      ],
      "id": "8bff8d06-64df-417f-a19c-5e720301c024"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# 1\u884c\u76ee\u306e\u30c7\u30fc\u30bf\u3092\u4f7f\u3063\u3066\u5217\u306e\u540d\u524d\u3092\u4ed8\u3051\u76f4\u3059\uff0e\n",
        "colnames(tmp) <- tmp[1, ]\n",
        "\n",
        "# 1\u884c\u76ee\u30681\u5217\u76ee\u3092\u524a\u9664\u3059\u308b\uff0e\n",
        "my_data <- tmp[-1, -1]\n",
        "my_data"
      ],
      "id": "115e5c02-6aae-46d7-81ba-61f2ebde2a20"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(jsonlite)\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/exam.json\")\n",
        "my_data <- fromJSON(my_url)\n",
        "#my_data <- fromJSON(\"exam.json\") # \uff08\u30d5\u30a1\u30a4\u30eb\u3092\u4f7f\u3046\u5834\u5408\uff09\n",
        "my_data"
      ],
      "id": "4cea1087-3681-4ab2-8c95-6fb5fe3d41f7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(xml2)\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/exam.xml\")\n",
        "my_xml <- read_xml(my_url)      # XML\u30c7\u30fc\u30bf\u306e\u8aad\u307f\u8fbc\u307f\n",
        "#my_xml <- read_xml(\"exam.xml\") # \uff08\u30d5\u30a1\u30a4\u30eb\u3092\u4f7f\u3046\u5834\u5408\uff09\n",
        "xml_ns(my_xml)                  # \u540d\u524d\u7a7a\u9593\u306e\u78ba\u8a8d\uff08d1\uff09"
      ],
      "id": "82d78d62-0849-4f72-b27c-b9ca6b68cd75"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_records <- xml_find_all(my_xml, \".//d1:record\")"
      ],
      "id": "89c48d3e-74ab-486d-94d7-85214097b7c0"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "f <- function(record) {\n",
        "  tmp <- xml_attrs(record)                    # \u5c5e\u6027\u3092\u5168\u3066\u53d6\u308a\u51fa\u3057\uff0c\n",
        "  xml_children(record) %>% walk(function(e) {\n",
        "    tmp[xml_name(e)] <<- xml_text(e)          # \u5b50\u8981\u7d20\u306e\u540d\u524d\u3068\u5185\u5bb9\u3092\u8ffd\u52a0\u3059\u308b\uff0e\n",
        "  })\n",
        "  tmp\n",
        "}"
      ],
      "id": "a6048b0e-89e9-4e26-b7eb-e9b106177b0b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_data <- my_records %>% map_dfr(f)\n",
        "my_data$english <- as.numeric(my_data$english)\n",
        "my_data$math    <- as.numeric(my_data$math)\n",
        "my_data"
      ],
      "id": "1da50026-073f-4153-8673-39b3a9dd1747"
    },
    {
      "id": "23841c33",
      "cell_type": "markdown",
      "source": "## 5.2 \u30c7\u30fc\u30bf\u306e\u5909\u63db",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x1 <- c(1, 2, 3)\n",
        "\n",
        "z1 <- scale(x1)\n",
        "# \u3042\u308b\u3044\u306f\n",
        "z1 <- (x1 - mean(x1)) / sd(x1)\n",
        "\n",
        "z1"
      ],
      "id": "744710ea-75b5-47c5-add6-82067a60d3cb"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "c(mean(z1), sd(z1))"
      ],
      "id": "758d60a8-9cb0-47e2-a1e8-1db96c145dca"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "z1 * sd(x1) + mean(x1)"
      ],
      "id": "f6499da0-080d-4009-acf0-ee499b422f12"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x2 <- c(1, 3, 5)\n",
        "z2 <- (x2 - mean(x1)) / sd(x1)\n",
        "c(mean(z2), sd(z2))"
      ],
      "id": "f8b569ca-ae77-496e-8394-75b15cd8c28b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "\n",
        "my_df <- data.frame(\n",
        "  id = c(1, 2, 3),\n",
        "  class = as.factor(\n",
        "    c(\"A\", \"B\", \"C\")))\n",
        "\n",
        "my_enc <- my_df %>%\n",
        "  dummyVars(formula = ~ .)\n",
        "\n",
        "my_enc %>% predict(my_df)"
      ],
      "id": "ea6a9608-e3fd-43cf-b458-e0a99aa1c01f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_df2 <- data.frame(\n",
        "  id =    c( 4 ,  5 ,  6 ),\n",
        "  class = c(\"B\", \"C\", \"B\"))\n",
        "my_enc %>% predict(my_df2)"
      ],
      "id": "41ec009c-6d88-4ded-afe0-198dd79ee151"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_enc <- my_df %>%\n",
        "  dummyVars(formula = ~ .,\n",
        "            fullRank = TRUE)\n",
        "my_enc %>% predict(my_df)\n",
        "\n",
        "my_enc %>% predict(my_df2)"
      ],
      "id": "51127bdc-1f06-4b4a-a3f8-c773d1975893"
    },
    {
      "id": "2107dacd",
      "cell_type": "markdown",
      "source": "# 6 \u6a5f\u68b0\u5b66\u7fd2\u306e\u76ee\u7684\u30fb\u30c7\u30fc\u30bf\u30fb\u624b\u6cd5\n\n\n",
      "metadata": {}
    },
    {
      "id": "4d8aac8b",
      "cell_type": "markdown",
      "source": "## 6.1 \u6a5f\u68b0\u5b66\u7fd2\u306e\u76ee\u7684\uff08\u672c\u66f8\u306e\u5834\u5408\uff09\n\n\n",
      "metadata": {}
    },
    {
      "id": "cd300d92",
      "cell_type": "markdown",
      "source": "## 6.2 \u6a5f\u68b0\u5b66\u7fd2\u306e\u305f\u3081\u306e\u30c7\u30fc\u30bf",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "iris\n",
        "# \u4ee5\u4e0b\u7701\u7565"
      ],
      "id": "a2be3226-8258-47d4-b236-6ead903be717"
    },
    {
      "id": "c0b32958",
      "cell_type": "markdown",
      "source": "## 6.3 \u6a5f\u68b0\u5b66\u7fd2\u306e\u305f\u3081\u306e\u624b\u6cd5\n\n\n",
      "metadata": {}
    },
    {
      "id": "477ba90d",
      "cell_type": "markdown",
      "source": "# 7 \u56de\u5e301\uff08\u5358\u56de\u5e30\uff09",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Google Colaboratory\u306e\u74b0\u5883\u8a2d\u5b9a\n",
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  options(Ncpus = parallel::detectCores())\n",
        "  installed_packages <- rownames(installed.packages())\n",
        "  packages_to_install <- c(\"caret\", \"doParallel\", \"pastecs\")\n",
        "  install.packages(setdiff(packages_to_install, installed_packages))\n",
        "}"
      ],
      "id": "48efd7fa-ebe6-48b0-8ff2-5c744ca394fc"
    },
    {
      "id": "f8e544e4",
      "cell_type": "markdown",
      "source": "## 7.1 \u81ea\u52d5\u8eca\u306e\u505c\u6b62\u8ddd\u96e2\n\n\n",
      "metadata": {}
    },
    {
      "id": "a93390b1",
      "cell_type": "markdown",
      "source": "## 7.2 \u30c7\u30fc\u30bf\u306e\u78ba\u8a8d",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- cars"
      ],
      "id": "9b196191-b7d4-4fe2-93c8-d007b4f02f2a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "dim(my_data)"
      ],
      "id": "723dfa95-d712-402f-9624-559c8c25c236"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "head(my_data)"
      ],
      "id": "e4f9d8ea-3f80-44d8-bbdf-adc59e6aa254"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "options(digits = 3)\n",
        "pastecs::stat.desc(my_data)"
      ],
      "id": "984355cf-560b-4d36-b634-43b572bd7648"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_data %>%\n",
        "  ggplot(aes(x = speed, y = dist)) +\n",
        "  geom_point()"
      ],
      "id": "54cff757-523e-4c6a-b6f2-6d421dcb2847"
    },
    {
      "id": "d9d5b808",
      "cell_type": "markdown",
      "source": "## 7.3 \u56de\u5e30\u5206\u6790",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "\n",
        "my_data <- cars\n",
        "tmp <- data.frame(speed = 21.5, dist = 67)\n",
        "my_data %>% ggplot(aes(x = speed, y = dist)) +\n",
        "  coord_cartesian(xlim = c(4, 25), ylim = c(0, 120)) +\n",
        "  geom_point() +\n",
        "  stat_smooth(formula = y ~ x, method = \"lm\") +\n",
        "  geom_pointrange(data = tmp, aes(ymin = -9, ymax = dist),  linetype = \"dotted\") +\n",
        "  geom_pointrange(data = tmp, aes(xmin =  0, xmax = speed), linetype = \"dotted\")"
      ],
      "id": "607b9529-e1f2-4d36-90fc-f368f188aab0"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- cars"
      ],
      "id": "4641fe2a-fac2-48dc-8bdb-1f44fa4cca81"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = dist ~ speed, # \u30e2\u30c7\u30eb\u5f0f\uff08\u51fa\u529b\u5909\u6570\u3068\u5165\u529b\u5909\u6570\u306e\u95a2\u4fc2\uff09\n",
        "                  data = my_data,      # \u30c7\u30fc\u30bf\n",
        "                  method = \"lm\")       # \u624b\u6cd5"
      ],
      "id": "a731a865-940e-4583-9b69-b374a8465adb"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "coef(my_model$finalModel)"
      ],
      "id": "9fb4ebb8-3fae-4296-800d-ec4d5d1b33f7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- data.frame(speed = 21.5)\n",
        "my_model %>% predict(tmp)"
      ],
      "id": "ff4a2ea1-0c30-4d93-a924-5d3a0552de89"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "f <- function(x) { my_model %>% predict(data.frame(speed = x)) }"
      ],
      "id": "51e1cc39-7a37-459e-aae4-6e24d7becfda"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_data %>%\n",
        "  ggplot(aes(x = speed, y = dist,\n",
        "             color = \"data\")) +\n",
        "  geom_point() +\n",
        "  stat_function(\n",
        "    fun = f,\n",
        "    mapping = aes(color = \"model\"))"
      ],
      "id": "e3038083-9763-4f35-a00c-95caa6046117"
    },
    {
      "id": "f63f7698",
      "cell_type": "markdown",
      "source": "## 7.4 \u5f53\u3066\u306f\u307e\u308a\u306e\u826f\u3055\u306e\u6307\u6a19",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- cars\n",
        "my_model <- train(form = dist ~ speed, data = my_data, method = \"lm\")\n",
        "\n",
        "y  <- my_data$dist\n",
        "y_ <- my_model %>% predict(my_data)\n",
        "my_data$y_ <- y_"
      ],
      "id": "2e75dd51-8cca-4be5-a368-99ad74df1882"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_data$residual <- y - y_\n",
        "head(my_data)"
      ],
      "id": "f8fb356e-4227-48c8-8c7c-202044595f01"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_data %>%\n",
        "  ggplot(aes(x = speed, y = dist)) +\n",
        "  geom_point() +\n",
        "  geom_line(aes(x = speed, y = y_)) +\n",
        "  geom_linerange(mapping = aes(ymin = y_, ymax = dist), linetype = \"dotted\")"
      ],
      "id": "0cf87571-65df-4318-80a1-b48fb8439643"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "RMSE(y_, y)\n",
        "# \u3042\u308b\u3044\u306f\n",
        "mean((my_data$residual^2))**0.5\n"
      ],
      "id": "d4e51d53-9e1a-4655-83dc-eff101a27c3e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "R2(pred = y_, obs = y,\n",
        "   form = \"traditional\")"
      ],
      "id": "8482915e-ede9-44c4-a06c-d3cfa408f1f9"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "R2(pred = y_, obs = y,\n",
        "   form = \"corr\")\n",
        "# \u3042\u308b\u3044\u306f\n",
        "summary(my_model$finalModel)$r.squared"
      ],
      "id": "d6aef545-431b-49a8-8a2c-19f2e9f365dc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_test <- my_data[1:3, ]\n",
        "y  <- my_test$dist\n",
        "y_ <- my_model %>% predict(my_test)\n",
        "\n",
        "R2(pred = y_, obs = y,\n",
        "   form = \"traditional\")\n",
        "\n",
        "R2(pred = y_, obs = y,\n",
        "   form = \"corr\")"
      ],
      "id": "e94f09e4-16ad-46c6-99e6-d76e04b6aa81"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- cars\n",
        "my_idx <- c(2, 11, 27, 34, 39, 44)\n",
        "my_sample <- my_data[my_idx, ]"
      ],
      "id": "b04fc193-8242-45be-a02c-282f5747ac95"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "options(warn = -1) # \u3053\u308c\u4ee5\u964d\uff0c\u8b66\u544a\u3092\u8868\u793a\u3057\u306a\u3044\uff0e\n",
        "my_model <- train(form = dist ~ poly(speed, degree = 5, raw = TRUE),\n",
        "                  data = my_sample,\n",
        "                  method = \"lm\")\n",
        "options(warn = 0)  # \u3053\u308c\u4ee5\u964d\uff0c\u8b66\u544a\u3092\u8868\u793a\u3059\u308b\uff0e\n",
        "\n",
        "y  <- my_sample$dist\n",
        "y_ <- my_model %>% predict(my_sample)"
      ],
      "id": "eddac552-073f-43c1-9f69-efa2c03c17c6"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "RMSE(y_, y)\n",
        "\n",
        "R2(pred = y_, obs = y,\n",
        "   form = \"traditional\")\n",
        "\n",
        "R2(pred = y_, obs = y,\n",
        "   form = \"corr\")"
      ],
      "id": "d48e8d85-693f-4c33-ae05-7c1b7e838ba4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "f <- function(x) { my_model %>% predict(data.frame(speed = x)) }\n",
        "\n",
        "my_data %>%\n",
        "  ggplot(aes(x = speed, y = dist, color = \"data\")) +\n",
        "  geom_point() +\n",
        "  geom_point(data = my_sample, mapping = aes(color = \"sample\")) +\n",
        "  stat_function(fun = f, mapping = aes(color = \"model\")) +\n",
        "  coord_cartesian(ylim = c(0, 120))"
      ],
      "id": "f9e87ecc-d2f3-4025-a1f0-f5a4077535a3"
    },
    {
      "id": "86c5a214",
      "cell_type": "markdown",
      "source": "## 7.5 K\u6700\u8fd1\u508d\u6cd5",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u6e96\u5099\n",
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- cars\n",
        "\n",
        "# \u8a13\u7df4\n",
        "my_model <- train(form = dist ~ speed, data = my_data, method = \"knn\")\n",
        "\n",
        "# \u53ef\u8996\u5316\u306e\u6e96\u5099\n",
        "f <- function(x) { my_model %>% predict(data.frame(speed = x))}"
      ],
      "id": "3c0ae9cd-f649-4add-84c3-e14143d56e3e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_data %>%\n",
        "  ggplot(aes(x = speed,\n",
        "             y = dist,\n",
        "             color = \"data\")) +\n",
        "  geom_point() +\n",
        "  stat_function(\n",
        "    fun = f,\n",
        "    mapping = aes(color = \"model\"))"
      ],
      "id": "74735d61-f9b7-4a8e-9e82-d90155075afd"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y  <- my_data$dist\n",
        "y_ <- my_model %>% predict(my_data)\n",
        "\n",
        "RMSE(y_, y)\n",
        "\n",
        "R2(pred = y_, obs = y,\n",
        "   form = \"traditional\")\n",
        "\n",
        "R2(pred = y_, obs = y,\n",
        "   form = \"corr\")"
      ],
      "id": "7ab8fbe5-19fe-498f-b6c5-d5a6d0f6b937"
    },
    {
      "id": "95c29cd4",
      "cell_type": "markdown",
      "source": "## 7.6 \u691c\u8a3c",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- cars\n",
        "my_model <- train(form = dist ~ speed, data = my_data, method = \"lm\")\n",
        "\n",
        "my_model$results"
      ],
      "id": "7126dd63-d822-40c7-8a83-d99551db203f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = dist ~ speed, data = my_data, method = \"lm\",\n",
        "                  trControl = trainControl(method = \"cv\", number = 5))\n",
        "my_model$results"
      ],
      "id": "6b3386bb-193d-4f02-b9f2-757b6d06f05e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = dist ~ speed, data = my_data, method = \"lm\",\n",
        "                  trControl = trainControl(method = \"LOOCV\"))\n",
        "my_model$results"
      ],
      "id": "679eeba2-a963-4ab6-8909-b8b32d5080d7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(doParallel)\n",
        "cl <- makeCluster(detectCores())\n",
        "registerDoParallel(cl)"
      ],
      "id": "5e817ce5-2ba7-4a24-83b9-6f3e3421f789"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- cars\n",
        "my_model <- train(form = dist ~ speed, data = my_data, method = \"lm\")\n",
        "y  <- my_data$dist\n",
        "y_ <- my_model %>% predict(my_data)"
      ],
      "id": "2ebe7790-88e0-4c9c-b3b6-413b45e1022a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# RMSE\uff08\u8a13\u7df4\uff09\n",
        "RMSE(y_, y)\n",
        "\n",
        "# \u6c7a\u5b9a\u4fc2\u65701\uff08\u8a13\u7df4\uff09\n",
        "R2(pred = y_, obs = y,\n",
        "   form = \"traditional\")\n",
        "\n",
        "# \u6c7a\u5b9a\u4fc2\u65706\uff08\u8a13\u7df4\uff09\n",
        "R2(pred = y_, obs = y,\n",
        "   form = \"corr\")"
      ],
      "id": "015e893e-ac2a-4d35-813d-2ddb68550cb7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "postResample(pred = y_, obs = y)"
      ],
      "id": "f69bf55f-5428-43c5-abb3-710a76047c16"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = dist ~ speed, data = my_data, method = \"lm\")\n",
        "my_model$results\n",
        "# \u5de6\u304b\u3089\uff0cRMSE\uff08\u691c\u8a3c\uff09\uff0c\u6c7a\u5b9a\u4fc2\u65706\uff08\u691c\u8a3c\uff09\uff0cMAE\uff08\u691c\u8a3c\uff09"
      ],
      "id": "7d65bf96-8602-4390-9594-4cb3352a47bd"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = dist ~ speed, data = my_data, method = \"lm\",\n",
        "                  trControl = trainControl(method = \"LOOCV\"))\n",
        "\n",
        "# \u65b9\u6cd51\n",
        "my_model$results\n",
        "\n",
        "# \u65b9\u6cd52\n",
        "y  <- my_model$pred$obs\n",
        "y_ <- my_model$pred$pred\n",
        "mean((y - y_)^2)**0.5"
      ],
      "id": "9f863a74-2baf-4f5d-a93c-49edcd08cd55"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "mean(((y - y_)^2)**0.5)"
      ],
      "id": "f253bd69-2cb5-45cf-9fe7-ef65a8ee1f2b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- cars\n",
        "\n",
        "my_lm_model <- train(form = dist ~ speed, data = my_data, method = \"lm\",\n",
        "                     trControl = trainControl(method = \"LOOCV\"))\n",
        "\n",
        "my_knn_model <- train(form = dist ~ speed, data = my_data, method = \"knn\",\n",
        "                      tuneGrid = data.frame(k = 5),\n",
        "                      trControl = trainControl(method = \"LOOCV\"))"
      ],
      "id": "e19a72fc-ecfa-41eb-8748-e6c60a452790"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_lm_model$results$RMSE\n",
        "\n",
        "my_knn_model$results$RMSE"
      ],
      "id": "344e1667-0c18-4efe-bd58-a4e7f4cc4ab0"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y     <- my_data$dist\n",
        "y_lm  <- my_lm_model$pred$pred\n",
        "y_knn <- my_knn_model$pred$pred\n",
        "\n",
        "my_df <- data.frame(\n",
        "  lm  = (y - y_lm)^2,\n",
        "  knn = (y - y_knn)^2)\n",
        "\n",
        "head(my_df)"
      ],
      "id": "22951462-27ff-40f2-b53b-0934a1d7ec7e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "boxplot(my_df, ylab = \"r^2\")"
      ],
      "id": "807dd781-3492-4f0f-badf-98cafd2f1a66"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "t.test(x = my_df$lm, y = my_df$knn,\n",
        "       conf.level = 0.95,\n",
        "       paired = TRUE,\n",
        "       alternative = \"two.sided\")\n"
      ],
      "id": "62676836-cd23-43aa-8434-3a710f24b6b8"
    },
    {
      "id": "62250f0c",
      "cell_type": "markdown",
      "source": "## 7.7 \u30d1\u30e9\u30e1\u30fc\u30bf\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- cars\n",
        "my_model <- train(form = dist ~ speed, data = my_data, method = \"knn\")\n",
        "my_model$results"
      ],
      "id": "a024695b-0c7a-4d51-9470-afbba9e93008"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_params <- expand.grid(k = 1:15)\n",
        "\n",
        "my_model <- train(form = dist ~ speed, data = my_data, method = \"knn\",\n",
        "                  tuneGrid = my_params,\n",
        "                  trControl = trainControl(method = \"LOOCV\"))"
      ],
      "id": "afe48cdb-25da-4007-a83e-16d89da10cbc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "head(my_model$results)"
      ],
      "id": "fb927952-6673-4292-8e55-4b234f983de6"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "ggplot(my_model)"
      ],
      "id": "87b75821-ece4-4804-8bf5-c8f2548fb040"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model$bestTune"
      ],
      "id": "34474e33-477f-4965-8fcc-1ba44ca64b56"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model$results %>%\n",
        "  filter(RMSE == min(RMSE))"
      ],
      "id": "00dbe773-659d-48c3-b970-ef7fcdf146e6"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y  <- my_data$dist\n",
        "y_ <- my_model %>% predict(my_data)\n",
        "RMSE(y_, y)"
      ],
      "id": "19f2d291-7d3d-4a20-9ab4-2a9fc65d9691"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- cars\n",
        "\n",
        "my_loocv <- function(k) {\n",
        "  my_model <- train(form = dist ~ speed, data = my_data, method = \"knn\",\n",
        "                    tuneGrid = data.frame(k = k),\n",
        "                    trControl = trainControl(method = \"LOOCV\"))\n",
        "  y  <- my_data$dist\n",
        "  y_ <- my_model %>% predict(my_data)\n",
        "  list(k = k,\n",
        "       training = RMSE(y_, y),             # RMSE\uff08\u8a13\u7df4\uff09\n",
        "       validation = my_model$results$RMSE) # RMSE\uff08\u691c\u8a3c\uff09\n",
        "}\n",
        "\n",
        "my_results <- 1:15 %>% map_dfr(my_loocv)"
      ],
      "id": "23e5b662-a152-45d0-8d43-924f83549afb"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_results %>%\n",
        "  pivot_longer(-k) %>%\n",
        "  ggplot(aes(x = k, y = value,\n",
        "             color = name)) +\n",
        "  geom_line() + geom_point() +\n",
        "  xlab(\"#Neighbors\") + ylab(\"RMSE\") +\n",
        "  theme(legend.position = c(1, 0),\n",
        "        legend.justification = c(1, 0))"
      ],
      "id": "8aa74542-792d-43e5-8d4b-bf2778eb00ce"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u4e26\u5217\u30af\u30e9\u30b9\u30bf\u306e\u505c\u6b62\n",
        "stopCluster(cl)\n",
        "registerDoSEQ()"
      ],
      "id": "ec6fd557-6dcb-4f18-a999-7c947576e7f7"
    },
    {
      "id": "9cc4a3c9",
      "cell_type": "markdown",
      "source": "# 8 \u56de\u5e302\uff08\u91cd\u56de\u5e30\uff09",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Google Colaboratory\u306e\u74b0\u5883\u8a2d\u5b9a\n",
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  options(Ncpus = parallel::detectCores())\n",
        "  installed_packages <- rownames(installed.packages())\n",
        "  packages_to_install <- c(\"caret\", \"ggfortify\", \"glmnetUtils\", \"leaps\", \"neuralnet\", \"psych\")\n",
        "  install.packages(setdiff(packages_to_install, installed_packages))\n",
        "}"
      ],
      "id": "646e31c5-9167-486f-9442-ae43fff6c4cf"
    },
    {
      "id": "9fd034f7",
      "cell_type": "markdown",
      "source": "## 8.1 \u30d6\u30c9\u30a6\u306e\u751f\u80b2\u6761\u4ef6\u3068\u30ef\u30a4\u30f3\u306e\u4fa1\u683c",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "my_url <- \"http://www.liquidasset.com/winedata.html\"\n",
        "tmp <- read.table(file = my_url,   # \u8aad\u307f\u8fbc\u3080\u5bfe\u8c61\n",
        "                  header = TRUE,   # 1\u884c\u76ee\u306f\u5909\u6570\u540d\n",
        "                  na.string = \".\", # \u6b20\u640d\u5024\u3092\u8868\u3059\u6587\u5b57\u5217\n",
        "                  skip = 62,       # \u8aad\u307f\u98db\u3070\u3059\u884c\u6570\n",
        "                  nrows = 38)      # \u8aad\u307f\u8fbc\u3080\u884c\u6570\n",
        "psych::describe(tmp)"
      ],
      "id": "e56f85eb-8189-4d8f-a4ee-d072687e404b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_data <- na.omit(tmp[, -c(1, 2)])\n",
        "head(my_data)"
      ],
      "id": "16bee85f-8a10-4c96-bc93-3cbfd86ed82d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "dim(my_data)"
      ],
      "id": "b461e3ce-7942-4122-b212-a74511e4ea44"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_data %>% write_csv(\"wine.csv\")"
      ],
      "id": "77054ded-1255-469c-978a-814736ff10f2"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "#my_data <- read_csv(\"wine.csv\") # \u4f5c\u3063\u305f\u30d5\u30a1\u30a4\u30eb\u3092\u4f7f\u3046\u5834\u5408\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/wine.csv\")\n",
        "my_data <- read_csv(my_url)"
      ],
      "id": "01f49778-181e-4623-aaeb-43574bf5a3fb"
    },
    {
      "id": "e283ed38",
      "cell_type": "markdown",
      "source": "## 8.2 \u91cd\u56de\u5e30\u5206\u6790",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/wine.csv\")\n",
        "my_data <- read_csv(my_url)\n",
        "\n",
        "my_model <- train(form = LPRICE2 ~ WRAIN + DEGREES + HRAIN + TIME_SV,\n",
        "                  data = my_data,\n",
        "                  method = \"lm\",\n",
        "                  trControl = trainControl(method = \"LOOCV\"))"
      ],
      "id": "2060cc9e-eaae-4f0a-a360-4dc700a2d22a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "coef(my_model$finalModel) %>%\n",
        "  as.data.frame"
      ],
      "id": "8ef22205-0463-4006-9b5e-5a0cf72f1ef0"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_test <- data.frame(\n",
        "  WRAIN = 500, DEGREES = 17,\n",
        "  HRAIN = 120, TIME_SV = 2)\n",
        "my_model %>% predict(my_test)"
      ],
      "id": "a24bc4da-2f4e-4b91-8ed7-40eaba6a8fe9"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y  <- my_data$LPRICE2\n",
        "y_ <- my_model %>% predict(my_data)\n",
        "\n",
        "RMSE(y_, y)\n",
        "\n",
        "R2(pred = y_, obs = y,\n",
        "   form = \"traditional\")\n",
        "\n",
        "R2(pred = y_, obs = y,\n",
        "   form = \"corr\")"
      ],
      "id": "16f10e75-788f-4931-9840-0b3dc333d1b5"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model$results"
      ],
      "id": "5519aa26-d550-4093-b7c2-6287d657e3d4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "M <- my_data[, -1] %>%\n",
        "  mutate(b0 = 1) %>% as.matrix\n",
        "b <- MASS::ginv(M) %*% y\n",
        "matrix(b,\n",
        "       dimnames = list(colnames(M)))"
      ],
      "id": "587dd50b-df55-49ba-bdda-07a7f7bf1ff2"
    },
    {
      "id": "ca7bc62a",
      "cell_type": "markdown",
      "source": "## 8.3 \u6a19\u6e96\u5316",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/wine.csv\")\n",
        "my_data <- read_csv(my_url)\n",
        "\n",
        "my_data %>%\n",
        "  mutate_if(is.numeric, scale) %>% # \u6570\u5024\u306e\u5217\u306e\u6a19\u6e96\u5316\n",
        "  pivot_longer(-LPRICE2) %>%\n",
        "  ggplot(aes(x = name, y = value)) +\n",
        "  geom_boxplot() +\n",
        "  stat_summary(fun = mean, geom = \"point\", size = 3) +\n",
        "  xlab(NULL)"
      ],
      "id": "0833716e-555d-456a-b184-024304a9bb36"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(\n",
        "  form = LPRICE2 ~ .,\n",
        "  data = my_data,\n",
        "  method = \"lm\",\n",
        "  preProcess = c(\"center\", \"scale\"))"
      ],
      "id": "851f5702-3095-4915-9fe9-8e85b1d2d98d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "coef(my_model$finalModel) %>%\n",
        "  as.data.frame"
      ],
      "id": "6086d010-4394-4b5f-b272-5a64a54f5b1b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_test <- data.frame(\n",
        "  WRAIN = 500, DEGREES = 17,\n",
        "  HRAIN = 120, TIME_SV = 2)\n",
        "my_model %>% predict(my_test)"
      ],
      "id": "b8f8c218-cc44-4517-b346-7465a638c3a1"
    },
    {
      "id": "e22dc4c3",
      "cell_type": "markdown",
      "source": "## 8.4 \u5165\u529b\u5909\u6570\u306e\u6570\u3068\u30e2\u30c7\u30eb\u306e\u826f\u3055",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/wine.csv\")\n",
        "my_data <- read_csv(my_url)\n",
        "\n",
        "n <- nrow(my_data)\n",
        "my_data2 <- my_data %>% mutate(v1 = 0:(n - 1) %% 2,\n",
        "                               v2 = 0:(n - 1) %% 3)\n",
        "head(my_data2)"
      ],
      "id": "fd4aefa2-e923-4731-95d9-e4f12f590741"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model2 <- train(form = LPRICE2 ~ ., data = my_data2, method = \"lm\",\n",
        "                   trControl = trainControl(method = \"LOOCV\"))\n",
        "y  <- my_data2$LPRICE2\n",
        "y_ <- my_model2 %>% predict(my_data2)\n",
        "\n",
        "RMSE(y_, y)\n",
        "\n",
        "my_model2$results$RMSE"
      ],
      "id": "e3557258-1cc4-4361-8ca5-889fce7ff342"
    },
    {
      "id": "b53ed4a8",
      "cell_type": "markdown",
      "source": "## 8.5 \u5909\u6570\u9078\u629e",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/wine.csv\")\n",
        "my_data <- read_csv(my_url)\n",
        "n <- nrow(my_data)\n",
        "my_data2 <- my_data %>% mutate(v1 = 0:(n - 1) %% 2,\n",
        "                               v2 = 0:(n - 1) %% 3)"
      ],
      "id": "1b81aa4c-2d11-4657-ad2d-de6ac0d04a26"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = LPRICE2 ~ .,\n",
        "                  data = my_data2,\n",
        "                  method = \"leapForward\", # \u5909\u6570\u5897\u52a0\u6cd5\n",
        "                  trControl = trainControl(method = \"LOOCV\"),\n",
        "                  tuneGrid = data.frame(nvmax = 1:6)) # \u9078\u629e\u3059\u308b\u5909\u6570\u306e\u4e0a\u9650\n",
        "summary(my_model$finalModel)$outmat"
      ],
      "id": "f6193301-49ad-40ec-aabd-6b3f0103254c"
    },
    {
      "id": "2d6eb607",
      "cell_type": "markdown",
      "source": "## 8.6 \u88dc\u8db3\uff1a\u6b63\u5247\u5316",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/wine.csv\")\n",
        "my_data <- read_csv(my_url)"
      ],
      "id": "b7e89f58-865c-45d8-9a8b-752427ea004c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "A <- 2\n",
        "B <- 0.1\n",
        "\n",
        "my_model <- train(\n",
        "  form = LPRICE2 ~ .,\n",
        "  data = my_data,\n",
        "  method = \"glmnet\",\n",
        "  standardize = TRUE,\n",
        "  tuneGrid = data.frame(\n",
        "    lambda = A,\n",
        "    alpha = B))"
      ],
      "id": "7cada87d-97a9-4ea7-9c47-b308931d1f66"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "coef(my_model$finalModel, A)"
      ],
      "id": "0b21ed10-7c97-4751-bf72-6ba798a68024"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_test <- data.frame(\n",
        "  WRAIN = 500, DEGREES = 17,\n",
        "  HRAIN = 120, TIME_SV = 2)\n",
        "my_model %>% predict(my_test)"
      ],
      "id": "dc8d12fb-1572-4c1e-ad92-a1db46ad07e5"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(ggfortify)\n",
        "library(glmnetUtils)\n",
        "\n",
        "my_data2 <- my_data %>% scale %>%\n",
        "  as.data.frame\n",
        "\n",
        "B <- 0.1\n",
        "\n",
        "glmnet(\n",
        "  form = LPRICE2 ~ .,\n",
        "  data = my_data2,\n",
        "  alpha = B) %>%\n",
        "  autoplot(xvar = \"lambda\") +\n",
        "  xlab(\"log A ( = log lambda)\") +\n",
        "  theme(legend.position = c(0.15, 0.25))"
      ],
      "id": "aa40d4ff-1cd2-4840-84a8-adfb8441649f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "As <- seq(0, 0.1, length.out = 21)\n",
        "Bs <- seq(0, 0.1, length.out =  6)\n",
        "\n",
        "my_model <- train(\n",
        "  form = LPRICE2 ~ ., data = my_data, method = \"glmnet\", standardize = TRUE,\n",
        "  trControl = trainControl(method = \"LOOCV\"),\n",
        "  tuneGrid = expand.grid(lambda = As, alpha  = Bs))\n",
        "\n",
        "my_model$bestTune"
      ],
      "id": "236a804a-7ffe-4cd6-94f2-c5b1ba8f14ef"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- \"B ( = alpha)\"\n",
        "ggplot(my_model) +\n",
        "  theme(legend.position = c(0, 1), legend.justification = c(0, 1)) +\n",
        "  xlab(\"A ( = lambda)\") +\n",
        "  guides(shape = guide_legend(tmp), color = guide_legend(tmp))"
      ],
      "id": "b5910514-d85d-4bd7-9403-25277fc03f09"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model$results %>%\n",
        "  filter(RMSE == min(RMSE))"
      ],
      "id": "31d21721-fff0-4887-b1cd-2740ea4a0d2c"
    },
    {
      "id": "8962cfb5",
      "cell_type": "markdown",
      "source": "## 8.7 \u30cb\u30e5\u30fc\u30e9\u30eb\u30cd\u30c3\u30c8\u30ef\u30fc\u30af",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "curve(1 / (1 + exp(-x)), -6, 6)"
      ],
      "id": "72c66cbc-542b-4f97-bb97-9b5ddc2e5bea"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/wine.csv\")\n",
        "my_data <- read_csv(my_url)"
      ],
      "id": "0e1566de-a41d-4367-8522-515ed8728aa5"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = LPRICE2 ~ .,\n",
        "                  data = my_data,\n",
        "                  method = \"neuralnet\",              # \u30cb\u30e5\u30fc\u30e9\u30eb\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\n",
        "                  preProcess = c(\"center\", \"scale\"), # \u6a19\u6e96\u5316\n",
        "                  trControl = trainControl(method = \"LOOCV\"))\n",
        "plot(my_model$finalModel) # \u8a13\u7df4\u6e08\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u306e\u63cf\u753b"
      ],
      "id": "41bfd981-9bfa-499c-a78f-061cce60a15e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model$results"
      ],
      "id": "03e66b77-09cc-4387-ba42-f69779b70e6b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(\n",
        "  form = LPRICE2 ~ .,\n",
        "  data = my_data,\n",
        "  method = \"neuralnet\",\n",
        "  preProcess = c(\"center\", \"scale\"),\n",
        "  trControl = trainControl(method = \"LOOCV\"),\n",
        "  tuneGrid = expand.grid(layer1 = 1:5,\n",
        "                         layer2 = 0:2,\n",
        "                         layer3 = 0))"
      ],
      "id": "860e22ea-6f69-4a65-bc53-2c0c62c6e57a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model$results %>%\n",
        "  filter(RMSE == min(RMSE))"
      ],
      "id": "8ab3a05a-889f-47ae-9477-0e27d7aed87b"
    },
    {
      "id": "6bc828b8",
      "cell_type": "markdown",
      "source": "# 9 \u5206\u985e1\uff08\u591a\u5024\u5206\u985e\uff09",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Google Colaboratory\u306e\u74b0\u5883\u8a2d\u5b9a\n",
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  options(Ncpus = parallel::detectCores())\n",
        "  installed_packages <- rownames(installed.packages())\n",
        "  packages_to_install <- c(\"caret\", \"furrr\", \"psych\", \"randomForest\", \"rpart.plot\", \"xgboost\")\n",
        "  install.packages(setdiff(packages_to_install, installed_packages))\n",
        "}"
      ],
      "id": "aed9a003-3fc0-4bf6-8a0c-4e92042bb2b8"
    },
    {
      "id": "06908ca2",
      "cell_type": "markdown",
      "source": "## 9.1 \u30a2\u30e4\u30e1\u306e\u30c7\u30fc\u30bf",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_data <- iris\n",
        "head(my_data)"
      ],
      "id": "0717a9e0-fdb2-4169-98cf-17a93d358474"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "psych::describe(my_data)"
      ],
      "id": "3c19e60d-b724-4f35-b664-55e09174e214"
    },
    {
      "id": "e29e40ce",
      "cell_type": "markdown",
      "source": "## 9.2 \u6728\u306b\u3088\u308b\u5206\u985e",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- iris\n",
        "my_model <- train(form = Species ~ ., data = my_data, method = \"rpart\")"
      ],
      "id": "823d3d39-42da-4393-bfaa-00f79ac34f56"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "rpart.plot::rpart.plot(my_model$finalModel, extra = 1)"
      ],
      "id": "4bb5c374-6da3-4d49-b78e-d082d992dfb2"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_test <- tribble(\n",
        "~Sepal.Length, ~Sepal.Width, ~Petal.Length, ~Petal.Width,\n",
        "          5.0,          3.5,           1.5,          0.5,\n",
        "          6.5,          3.0,           5.0,          2.0)\n",
        "\n",
        "my_model %>% predict(my_test)"
      ],
      "id": "72c41a3f-edf5-4634-b27f-eeef603ad736"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model %>% predict(my_test,\n",
        "                     type = \"prob\")"
      ],
      "id": "6297228c-8e18-4b1b-8bdb-8e4816db0f99"
    },
    {
      "id": "85fc4ffa",
      "cell_type": "markdown",
      "source": "## 9.3 \u6b63\u89e3\u7387",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- iris\n",
        "my_model <- train(form = Species ~ ., data = my_data, method = \"rpart2\")\n",
        "\n",
        "y  <- my_data$Species\n",
        "y_ <- my_model %>% predict(my_data)\n",
        "confusionMatrix(data = y_, reference = y)\n",
        "# \u4ee5\u4e0b\u306f\u5272\u611b"
      ],
      "id": "b088c3a4-cd75-4671-97f3-06c444f59bb0"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y  <- my_data$Species\n",
        "y_ <- my_model %>% predict(my_data)\n",
        "mean(y_ == y)"
      ],
      "id": "bfe0c91b-dd36-4ce0-878f-ad2421b37526"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = Species ~ ., data = my_data, method = \"rpart2\",\n",
        "                  trControl = trainControl(method = \"LOOCV\"))\n",
        "my_model$results"
      ],
      "id": "be743774-30d9-461f-9e51-dc73a787c147"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = Species ~ ., data = my_data, method = \"rpart2\",\n",
        "                  tuneGrid = data.frame(maxdepth = 1:10),\n",
        "                  trControl = trainControl(method = \"LOOCV\"))\n",
        "my_model$results %>% filter(Accuracy == max(Accuracy))"
      ],
      "id": "12e172c7-88ff-43e1-8be9-9480fe811d9c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u30d1\u30e9\u30e1\u30fc\u30bf\u3092\u4e0e\u3048\u308b\u3068\u6b63\u89e3\u7387\uff08LOOCV\uff09\u3092\u8fd4\u3059\u95a2\u6570\n",
        "my_loocv <- function(maxdepth, minbucket, minsplit) {\n",
        "  my_model <- train(form = Species ~ ., data = my_data, method = \"rpart2\",\n",
        "                    trControl = trainControl(method = \"LOOCV\"),\n",
        "                    tuneGrid = data.frame(maxdepth = maxdepth),\n",
        "                    control = rpart::rpart.control(cp = 0.01,\n",
        "                                                   minbucket = minbucket,\n",
        "                                                   minsplit = minsplit))\n",
        "  list(maxdepth = maxdepth,\n",
        "       minbucket = minbucket,\n",
        "       minsplit = minsplit,\n",
        "       Accuracy = my_model$results$Accuracy)\n",
        "}"
      ],
      "id": "3439870e-8614-4c13-aed3-c4a8e493671a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_params <- expand.grid(\n",
        "  maxdepth = 2:5,\n",
        "  minbucket = 1:7,\n",
        "  minsplit = c(2, 20))\n",
        "\n",
        "library(furrr)\n",
        "plan(multisession) # \u4e26\u5217\u51e6\u7406\u306e\u6e96\u5099\n",
        "my_results <- my_params %>% future_pmap_dfr(my_loocv, # \u5b9f\u884c\n",
        "  .options = furrr_options(seed = TRUE))\n",
        "\n",
        "my_results %>% filter(Accuracy == max(Accuracy)) # \u6b63\u89e3\u7387\uff08\u691c\u8a3c\uff09\u306e\u6700\u5927\u5024"
      ],
      "id": "ce8e40f1-16ee-445b-883b-ef7b22d2d167"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = Species ~ ., data = my_data, method = \"rpart2\",\n",
        "                  trControl = trainControl(method = \"none\"),\n",
        "                  tuneGrid = data.frame(maxdepth = 3),\n",
        "                  control = rpart::rpart.control(cp = 0.01,\n",
        "                                                 minbucket = 5,\n",
        "                                                 minsplit = 2))"
      ],
      "id": "05e4c3e4-fe0c-4d89-9d68-7a238bffc236"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "rpart.plot::rpart.plot(\n",
        "  my_model$finalModel, extra = 1)"
      ],
      "id": "c3c9f9e4-e434-4cb3-85db-0827e71cbdf4"
    },
    {
      "id": "7cdd29f5",
      "cell_type": "markdown",
      "source": "## 9.4 \u8907\u6570\u306e\u6728\u3092\u4f7f\u3046\u65b9\u6cd5",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- iris\n",
        "\n",
        "my_model <- train(form = Species ~ ., data = my_data, method = \"rf\",\n",
        "                  tuneGrid = data.frame(mtry = 2:4), # \u7701\u7565\u53ef\n",
        "                  trControl = trainControl(method = \"LOOCV\"))\n",
        "my_model$results"
      ],
      "id": "e5ae4082-a32b-42e3-a9a5-c731414ad2be"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(\n",
        "  form = Species ~ ., data = my_data, method = \"xgbTree\",\n",
        "  tuneGrid = expand.grid(\n",
        "    nrounds          = c(50, 100, 150),\n",
        "    max_depth        = c(1, 2, 3),\n",
        "    eta              = c(0.3, 0.4),\n",
        "    gamma            = 0,\n",
        "    colsample_bytree = c(0.6, 0.8),\n",
        "    min_child_weight = 1,\n",
        "    subsample        = c(0.5, 0.75, 1)),\n",
        "  trControl = trainControl(method = \"cv\", number = 5)) # 5\u5206\u5272\u4ea4\u5dee\u691c\u8a3c\n",
        "my_model$results %>% filter(Accuracy == max(Accuracy)) %>% head(5) %>% t"
      ],
      "id": "97e6d661-5595-4e9e-b65c-d2a4925711a7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = Species ~ ., data = my_data, method = \"rf\")\n",
        "ggplot(varImp(my_model))"
      ],
      "id": "ee512fc4-ab25-43e0-8736-1e21ab6b9e19"
    },
    {
      "id": "b59935ea",
      "cell_type": "markdown",
      "source": "## 9.5 \u6b20\u640d\u306e\u3042\u308b\u30c7\u30fc\u30bf\u3067\u306e\u5b66\u7fd2",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- iris\n",
        "\n",
        "n <- nrow(my_data)\n",
        "my_data$Petal.Length[0:(n - 1) %% 10 == 0] <- NA\n",
        "my_data$Petal.Width[ 0:(n - 1) %% 10 == 1] <- NA\n",
        "\n",
        "psych::describe(my_data) # n\u306e\u5024\u304c135\u306e\u5909\u6570\u306b\uff0c150 - 135 = 15\u500b\u306e\u6b20\u640d\u304c\u3042\u308b\uff0e"
      ],
      "id": "c96dca58-ab1e-4195-bcc1-601559dfbc2e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(\n",
        "  form = Species ~ ., data = my_data, method = \"rpart2\",\n",
        "  na.action = na.pass,         # \u6b20\u640d\u304c\u3042\u3063\u3066\u3082\u5b66\u7fd2\u3092\u6b62\u3081\u306a\u3044\uff0e\n",
        "  preProcess = \"medianImpute\", # \u6b20\u640d\u3092\u4e2d\u592e\u5024\u3067\u57cb\u3081\u308b\uff0e\n",
        "  trControl = trainControl(method = \"LOOCV\"),\n",
        "  tuneGrid = data.frame(maxdepth = 20),          # \u6728\u306e\u9ad8\u3055\u306e\u4e0a\u9650\n",
        "  control = rpart::rpart.control(minsplit = 2,   # \u5206\u5c90\u306e\u6761\u4ef6\n",
        "                                 minbucket = 1)) # \u7d42\u7aef\u30ce\u30fc\u30c9\u306e\u6761\u4ef6\n",
        "max(my_model$results$Accuracy)"
      ],
      "id": "e4305260-8dfd-4726-bf64-e8cbdb74d0b4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = Species ~ ., data = my_data, method = \"xgbTree\",\n",
        "                  na.action = na.pass,\n",
        "                  trControl = trainControl(method = \"cv\", number = 5))\n",
        "max(my_model$results$Accuracy)"
      ],
      "id": "39465e90-5a5d-4091-a3af-5f2b5a90fc54"
    },
    {
      "id": "ae3e38f1",
      "cell_type": "markdown",
      "source": "## 9.6 \u4ed6\u306e\u5206\u985e\u624b\u6cd5",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- iris\n",
        "\n",
        "my_model <- train(form = Species ~ ., data = my_data, method = \"knn\",\n",
        "                  trControl = trainControl(method = \"LOOCV\"))\n",
        "my_model$results %>% filter(Accuracy == max(Accuracy))"
      ],
      "id": "889f4226-c1fa-48b5-ad55-d77f8d3fe117"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(tidyverse)\n",
        "my_data <- iris\n",
        "\n",
        "my_model <- train(form = Species ~ ., data = my_data, method = \"nnet\",\n",
        "                  preProcess = c(\"center\", \"scale\"), # \u6a19\u6e96\u5316\n",
        "                  trControl = trainControl(method = \"LOOCV\"),\n",
        "                  trace = FALSE) # \u9014\u4e2d\u7d4c\u904e\u3092\u8868\u793a\u3057\u306a\u3044\n",
        "my_model$results %>% filter(Accuracy == max(Accuracy))"
      ],
      "id": "45790805-3116-4e08-a64d-ede1a19857ba"
    },
    {
      "id": "97bdb95f",
      "cell_type": "markdown",
      "source": "# 10 \u5206\u985e2\uff082\u5024\u5206\u985e\uff09",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Google Colaboratory\u306e\u74b0\u5883\u8a2d\u5b9a\n",
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  options(Ncpus = parallel::detectCores())\n",
        "  installed_packages <- rownames(installed.packages())\n",
        "  packages_to_install <- c(\"caret\", \"PRROC\", \"rpart.plot\")\n",
        "  install.packages(setdiff(packages_to_install, installed_packages))\n",
        "}"
      ],
      "id": "6e71f902-8be2-442a-880b-28f7f102dfac"
    },
    {
      "id": "2103b4e4",
      "cell_type": "markdown",
      "source": "## 10.1 2\u5024\u5206\u985e\u306e\u6027\u80fd\u6307\u6a19",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y       <- c(  0,   1,   1,   0,   1,   0,    1,   0,   0,   1)\n",
        "y_score <- c(0.7, 0.8, 0.3, 0.4, 0.9, 0.6, 0.99, 0.1, 0.2, 0.5)"
      ],
      "id": "cd9dbcc1-b8ce-4ca3-9cc4-7e86cc49b9da"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y_ <- ifelse(0.5 <= y_score, 1, 0)\n",
        "y_"
      ],
      "id": "e6a84b94-d1bf-4513-89ac-5906ad9582d5"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "confusionMatrix(data      = as.factor(y_), # \u4e88\u6e2c\n",
        "                reference = as.factor(y),  # \u6b63\u89e3\n",
        "                positive = \"1\",            # \u300c1\u300d\u3092\u967d\u6027\u3068\u3059\u308b\uff0e\n",
        "                mode = \"everything\")       # \u5168\u3066\u306e\u6307\u6a19\u3092\u6c42\u3081\u308b\uff0e"
      ],
      "id": "515aa260-60fc-4b86-a511-18800bc1484b"
    },
    {
      "id": "4747e755",
      "cell_type": "markdown",
      "source": "## 10.2 \u30c8\u30ec\u30fc\u30c9\u30aa\u30d5",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(PRROC)\n",
        "library(tidyverse)\n",
        "\n",
        "y       <- c(  0,   1,   1,   0,   1,   0,    1,   0,   0,   1)\n",
        "y_score <- c(0.7, 0.8, 0.3, 0.4, 0.9, 0.6, 0.99, 0.1, 0.2, 0.5)\n",
        "y_      <- ifelse(0.5 <= y_score, 1, 0)\n",
        "\n",
        "c(sum((y == 0) & (y_ == 1)) / sum(y == 0), # FPR\n",
        "  sum((y == 1) & (y_ == 1)) / sum(y == 1)) # TPR"
      ],
      "id": "36f0d2af-6f14-4de5-9db3-a27e4f12cef6"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_roc <- roc.curve(scores.class0 = y_score[y == 1], # \u7b54\u3048\u304c1\u306e\u3082\u306e\u306e\u30b9\u30b3\u30a2\n",
        "                    scores.class1 = y_score[y == 0], # \u7b54\u3048\u304c0\u306e\u3082\u306e\u306e\u30b9\u30b3\u30a2\n",
        "                    curve = TRUE)\n",
        "my_roc %>% plot(xlab = \"False Positive Rate\",\n",
        "                ylab = \"True Positive Rate\",\n",
        "                legend = FALSE)"
      ],
      "id": "d50d44a6-3118-47fa-8583-ca1c16acbd4c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_roc$auc"
      ],
      "id": "2a3a576a-baaa-4c05-bec3-50ed263cc9f3"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "c(sum((y == 1) & (y_ == 1)) / sum(y  == 1), # Recall == TPR\n",
        "  sum((y == 1) & (y_ == 1)) / sum(y_ == 1)) # Precision"
      ],
      "id": "ac0b4d37-daee-4fbf-9379-3189dc6e50a3"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_pr <- pr.curve(scores.class0 = y_score[y == 1],\n",
        "                  scores.class1 = y_score[y == 0],\n",
        "                  curve = TRUE)\n",
        "my_pr %>% plot(xlab = \"Recall\",\n",
        "               ylab = \"Precision\",\n",
        "               legend = FALSE)"
      ],
      "id": "e78152b5-2132-492d-9c52-9aad7766bea3"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_pr$auc.integral"
      ],
      "id": "06732dc1-f012-4fdd-9707-d83ebdf2cf33"
    },
    {
      "id": "a09d8f36",
      "cell_type": "markdown",
      "source": "## 10.3 \u30bf\u30a4\u30bf\u30cb\u30c3\u30af",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(PRROC)\n",
        "library(tidyverse)\n",
        "\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/titanic.csv\")\n",
        "my_data <- read_csv(my_url)"
      ],
      "id": "0adb3430-761b-4bf0-8ffb-922e311055bb"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "head(my_data)"
      ],
      "id": "9cd9a27b-1903-42ea-9139-e7053313f93e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- train(form = Survived ~ ., data = my_data, method = \"rpart2\",\n",
        "                  tuneGrid = data.frame(maxdepth = 2),\n",
        "                  trControl = trainControl(method = \"LOOCV\"))"
      ],
      "id": "9ced2bfe-58cf-4ef8-a0c3-b3a8a73811da"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "rpart.plot::rpart.plot(my_model$finalModel, extra = 1)"
      ],
      "id": "fec659cd-6a69-48bb-82ac-16220ba2b7fd"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model$results"
      ],
      "id": "b19bd5b2-5036-44e9-9d12-cd966f4b88dc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y <- my_data$Survived\n",
        "tmp <- my_model %>% predict(newdata = my_data, type = \"prob\")\n",
        "y_score <- tmp$Yes\n",
        "\n",
        "my_roc <- roc.curve(scores.class0 = y_score[y == \"Yes\"],\n",
        "                    scores.class1 = y_score[y == \"No\"],\n",
        "                    curve = TRUE)\n",
        "my_roc$auc\n",
        "\n",
        "my_roc %>% plot(xlab = \"False Positive Rate\",\n",
        "                ylab = \"True Positive Rate\",\n",
        "                legend = FALSE)"
      ],
      "id": "e0f2468a-75eb-4bc9-90aa-49e01ab4efd1"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "X <- my_data %>% select(Class) # \u8cea\u7684\u5165\u529b\u5909\u6570\n",
        "y <- my_data$Survived          # \u51fa\u529b\u5909\u6570\n",
        "\n",
        "options(warn = -1) # \u3053\u308c\u4ee5\u964d\uff0c\u8b66\u544a\u3092\u8868\u793a\u3057\u306a\u3044\uff0e\n",
        "my_model1 <- train(x = X, y = y, method = \"rpart2\",\n",
        "                   tuneGrid = data.frame(maxdepth = 2),\n",
        "                   trControl = trainControl(method = \"LOOCV\"))\n",
        "options(warn = 0)  # \u3053\u308c\u4ee5\u964d\uff0c\u8b66\u544a\u3092\u8868\u793a\u3059\u308b\uff0e\n",
        "\n",
        "rpart.plot::rpart.plot(my_model1$finalModel, extra = 1)\n",
        "my_model1$results"
      ],
      "id": "353f0d53-3c03-4f8f-882f-d6e48b6f72a5"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_enc <- my_data %>% dummyVars(formula = Survived ~ Class)\n",
        "my_data2 <- my_enc %>%\n",
        "  predict(my_data) %>%\n",
        "  as.data.frame %>%\n",
        "  mutate(Survived = my_data$Survived)\n",
        "\n",
        "my_model2 <- train(form = Survived ~ ., data = my_data2, method = \"rpart2\",\n",
        "                   tuneGrid = data.frame(maxdepth = 2),\n",
        "                   trControl = trainControl(method = \"LOOCV\"))\n",
        "rpart.plot::rpart.plot(my_model2$finalModel, extra = 1)\n",
        "my_model2$results"
      ],
      "id": "126778d7-16cf-4c06-87a7-708f9fe37694"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model3 <- train(form = Survived ~ Class, data = my_data, method = \"rpart2\",\n",
        "                   tuneGrid = data.frame(maxdepth = 2),\n",
        "                   trControl = trainControl(method = \"LOOCV\"))\n",
        "rpart.plot::rpart.plot(my_model3$finalModel, extra = 1)\n",
        "my_model3$results"
      ],
      "id": "03dd616e-b3dd-42de-876b-706b1b8994f0"
    },
    {
      "id": "34512698",
      "cell_type": "markdown",
      "source": "## 10.4 \u30ed\u30b8\u30b9\u30c6\u30a3\u30c3\u30af\u56de\u5e30",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "curve(1 / (1 + exp(-x)), -6, 6)"
      ],
      "id": "e7c64a3e-04ad-4f53-bbde-c13014682892"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "library(PRROC)\n",
        "library(tidyverse)\n",
        "\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/titanic.csv\")\n",
        "my_data <- read_csv(my_url)\n",
        "\n",
        "my_model <- train(form = Survived ~ ., data = my_data, method = \"glm\",\n",
        "                  trControl = trainControl(method = \"LOOCV\"))"
      ],
      "id": "32b9c246-54f9-462a-b217-f9a6af8c9b3f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "coef(my_model$finalModel) %>%\n",
        "  as.data.frame"
      ],
      "id": "a3706f49-7fc7-4cd1-ac19-6c991ce7c00a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model$results"
      ],
      "id": "5f4532d2-f2a8-4bc9-9628-e58d58597f82"
    },
    {
      "id": "214f11f6",
      "cell_type": "markdown",
      "source": "# 11 \u6df1\u5c64\u5b66\u7fd2\u3068AutoML",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Google Colaboratory\u306e\u74b0\u5883\u8a2d\u5b9a\n",
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  options(Ncpus = parallel::detectCores())\n",
        "  installed_packages <- rownames(installed.packages())\n",
        "  packages_to_install <- c(\"h2o\", \"keras3\")\n",
        "  install.packages(setdiff(packages_to_install, installed_packages))\n",
        "}"
      ],
      "id": "6f818bfe-817e-44c3-9144-a0d5d172f855"
    },
    {
      "id": "ccddbcb8",
      "cell_type": "markdown",
      "source": "## 11.1 Keras\u306b\u3088\u308b\u56de\u5e30",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  library(keras3)\n",
        "} else {\n",
        "  library(keras)\n",
        "}\n",
        "library(tidyverse)\n",
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/wine.csv\")\n",
        "tmp <- read_csv(my_url)"
      ],
      "id": "dfda32bf-1dac-4411-95a8-03ab0e89e74a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_data <- tmp[sample(nrow(tmp)), ]"
      ],
      "id": "4adb81b1-ea6b-4c52-b117-da671ffed6d2"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "X <- my_data %>%\n",
        "  select(-LPRICE2) %>% scale\n",
        "y <- my_data$LPRICE2"
      ],
      "id": "ef22a97d-0adf-494e-adff-bb92759b5252"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "curve(activation_relu(x), -3, 3)"
      ],
      "id": "f2da3268-5755-4b6d-bd54-fac1a3810541"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- keras_model_sequential() %>%\n",
        "  layer_dense(units = 3, activation = \"relu\", input_shape = c(4)) %>%\n",
        "  layer_dense(units = 1)\n",
        "\n",
        "summary(my_model) # \u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u306e\u6982\u8981\n",
        "# \u5272\u611b\uff08Python\u306e\u7d50\u679c\u3092\u53c2\u7167\uff09"
      ],
      "id": "72cd2cfb-2958-42e3-ae34-635f574464bb"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model %>% compile(\n",
        "  loss = \"mse\",\n",
        "  optimizer = \"rmsprop\")"
      ],
      "id": "622b7ba3-a871-4997-9586-85036eea2309"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_cb <- callback_early_stopping(\n",
        "  patience = 20,\n",
        "  restore_best_weights = TRUE)"
      ],
      "id": "f83317f7-c685-4655-8a2e-e6611646ce86"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_history <- my_model %>% fit(\n",
        "    x = X,\n",
        "    y = y,\n",
        "    validation_split = 0.25,\n",
        "    batch_size = 10,\n",
        "    epochs = 500,\n",
        "    callbacks = list(my_cb),\n",
        "    verbose = 0)"
      ],
      "id": "452a8626-84a8-49e2-a7ef-b7bd421ad2c7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "plot(my_history)"
      ],
      "id": "f15d19b9-59cc-4757-bd90-54de80e99ebc"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_history"
      ],
      "id": "6448eece-39a2-451f-a340-679e04c688f3"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y_ <- my_model %>% predict(X)\n",
        "mean((y_ - y)^2)**0.5"
      ],
      "id": "2d850414-261c-4cfb-a622-a7276f531286"
    },
    {
      "id": "636fe9d0",
      "cell_type": "markdown",
      "source": "## 11.2 Keras\u306b\u3088\u308b\u5206\u985e",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  library(keras3)\n",
        "} else {\n",
        "  library(keras)\n",
        "}\n",
        "library(tidyverse)\n",
        "my_data <- iris[sample(nrow(iris)), ]"
      ],
      "id": "81f5b906-c1c6-47c1-a90e-faa4ca25635e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "X <- my_data %>%\n",
        "  select(-Species) %>% scale\n",
        "y <- as.integer(my_data$Species) - 1"
      ],
      "id": "5c771f9d-684e-418f-8cd6-6bc03938f0e8"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- keras_model_sequential() %>%\n",
        "  layer_dense(units = 3, activation = \"relu\", input_shape = c(4)) %>%\n",
        "  layer_dense(units = 3, activation = \"softmax\")"
      ],
      "id": "85811335-0d20-4537-845f-75ceab9d63ad"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model %>% compile(loss = \"sparse_categorical_crossentropy\",\n",
        "                     optimizer = \"rmsprop\",\n",
        "                     metrics = c(\"accuracy\"))"
      ],
      "id": "5139693b-b061-4ff2-b5fe-e8069040a149"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_cb <- callback_early_stopping(\n",
        "  patience = 20,\n",
        "  restore_best_weights = TRUE)\n",
        "\n",
        "my_history <- my_model %>%\n",
        "  fit(x = X,\n",
        "    y = y,\n",
        "    validation_split = 0.25,\n",
        "    batch_size = 10,\n",
        "    epochs = 500,\n",
        "    callbacks = list(my_cb),\n",
        "    verbose = 0)\n",
        "\n",
        "plot(my_history)"
      ],
      "id": "5a07a2b2-daec-4544-b89d-6420c413503d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_history"
      ],
      "id": "5a36fbbf-8987-48cf-a303-82c73c47d13c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- my_model %>% predict(X)\n",
        "y_ <- apply(tmp, 1, which.max) - 1\n",
        "mean(y_ == y)"
      ],
      "id": "30fe2aac-d6f6-4d57-a708-ac29c5f97e2b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "-mean(log(c(0.8, 0.7, 0.3, 0.8)))\n",
        "\n",
        "-mean(log(c(0.7, 0.6, 0.2, 0.7)))"
      ],
      "id": "c8675cf3-62ab-4509-b9ce-2c330840e3d1"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y <- c(2, 1, 0, 1)\n",
        "y_1 <- rbind(c(0.1, 0.1, 0.8),\n",
        "             c(0.1, 0.7, 0.2),\n",
        "             c(0.3, 0.4, 0.3),\n",
        "             c(0.1, 0.8, 0.1))\n",
        "y_2 <- rbind(c(0.1, 0.2, 0.7),\n",
        "             c(0.2, 0.6, 0.2),\n",
        "             c(0.2, 0.5, 0.3),\n",
        "             c(0.2, 0.7, 0.1))"
      ],
      "id": "53ad4534-464b-4116-92af-7a2074ed14d1"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "c(mean(as.array(loss_sparse_categorical_crossentropy(y_true = y, y_pred = y_1))),\n",
        "  mean(as.array(loss_sparse_categorical_crossentropy(y_true = y, y_pred = y_2))))"
      ],
      "id": "9b6ad90a-9842-46da-8f3a-afb82aaad3ba"
    },
    {
      "id": "dde974a8",
      "cell_type": "markdown",
      "source": "## 11.3 MNIST\uff1a\u624b\u66f8\u304d\u6570\u5b57\u306e\u5206\u985e",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  library(keras3)\n",
        "} else {\n",
        "  library(keras)\n",
        "}\n",
        "library(tidyverse)\n",
        "c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()"
      ],
      "id": "89ab519f-f1d4-4031-aa8d-5ceb644d2d2b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "dim(x_train)"
      ],
      "id": "922f8f54-b3c6-4d8b-84d3-6393af500af4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x_train[5, , ]"
      ],
      "id": "e0878b26-11dd-4a38-b3af-a86dc15da918"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "plot(as.raster(x = x_train[5, , ],\n",
        "               max = max(x_train)))"
      ],
      "id": "beb08eed-34c0-43d2-9207-237059397062"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "head(y_train)"
      ],
      "id": "f4f3ff31-82f8-4c1a-8fcd-104bfb205a2d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "c(min(x_train), max(x_train))"
      ],
      "id": "ddaa36f1-3af4-407d-b704-3c47e62425e5"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x_train <- x_train / 255\n",
        "x_test  <- x_test  / 255"
      ],
      "id": "388cdf4b-1b1c-4043-b781-ab85e392171e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_index <- sample(1:60000, 6000)\n",
        "x_train <- x_train[my_index, , ]\n",
        "y_train <- y_train[my_index]"
      ],
      "id": "e8f1f0b2-6b7b-4df3-a914-dadd2bc37624"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- keras_model_sequential() %>%\n",
        "  layer_flatten(input_shape = c(28, 28)) %>%\n",
        "  layer_dense(units = 256, activation = \"relu\") %>%\n",
        "  layer_dense(units = 10, activation = \"softmax\")\n",
        "summary(my_model)\n",
        "# \u5272\u611b\uff08Python\u306e\u7d50\u679c\u3092\u53c2\u7167\uff09\n",
        "\n",
        "my_model %>% compile(loss = \"sparse_categorical_crossentropy\",\n",
        "                     optimizer = \"rmsprop\",\n",
        "                     metrics = c(\"accuracy\"))\n",
        "\n",
        "my_cb <- callback_early_stopping(patience = 5,\n",
        "                                 restore_best_weights = TRUE)"
      ],
      "id": "a1ebe4d5-1c58-4ff5-a4fa-c71f1199acb6"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_history <- my_model %>%\n",
        "  fit(x = x_train,\n",
        "      y = y_train,\n",
        "      validation_split = 0.2,\n",
        "      batch_size = 128,\n",
        "      epochs = 20,\n",
        "      callbacks = list(my_cb),\n",
        "      verbose = 0)\n",
        "\n",
        "plot(my_history)"
      ],
      "id": "9fdb87df-1e68-48ac-ab07-4aaa4d9ab00b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- my_model %>% predict(x_test)\n",
        "y_ <- apply(tmp, 1, which.max) - 1\n",
        "table(y_, y_test)"
      ],
      "id": "132f656a-1073-458e-84df-64a6f02b5f06"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [],
      "id": "abe93534-30d0-4e28-a57d-3b09dd959251"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "mean(y_ == y_test)"
      ],
      "id": "ae75042b-b3d9-47b7-9b62-9dfeb63cc965"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model %>%\n",
        "  evaluate(x = x_test, y = y_test)"
      ],
      "id": "4549dd35-898e-4259-a9b3-e3a405293be5"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "x_train2d <- x_train %>% array_reshape(c(-1, 28, 28, 1))\n",
        "x_test2d  <- x_test  %>% array_reshape(c(-1, 28, 28, 1))"
      ],
      "id": "ce72a21c-090a-45be-a007-56f6c5823cec"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- keras_model_sequential() %>%\n",
        "  layer_conv_2d(filters = 32, kernel_size = 3,  # \u7573\u307f\u8fbc\u307f\u5c64\n",
        "                activation = \"relu\",\n",
        "                input_shape = c(28, 28, 1)) %>%\n",
        "  layer_max_pooling_2d(pool_size = 2) %>%       # \u30d7\u30fc\u30ea\u30f3\u30b0\u5c64\n",
        "  layer_flatten() %>%\n",
        "  layer_dense(units = 128, activation = \"relu\") %>%\n",
        "  layer_dense(units = 10, activation = \"softmax\")\n",
        "\n",
        "summary(my_model)\n",
        "# \u5272\u611b\uff08Python \u306e\u7d50\u679c\u3092\u53c2\u7167\uff09\n",
        "\n",
        "my_model %>% compile(loss = \"sparse_categorical_crossentropy\",\n",
        "                     optimizer = \"rmsprop\",\n",
        "                     metrics = c(\"accuracy\"))\n",
        "\n",
        "my_cb <- callback_early_stopping(patience = 5,\n",
        "                                 restore_best_weights = TRUE)"
      ],
      "id": "a56b0e6b-64cd-47a0-a10a-686a75717b75"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_history <- my_model %>%\n",
        "  fit(x = x_train2d,\n",
        "      y = y_train,\n",
        "      validation_split = 0.2,\n",
        "      batch_size = 128,\n",
        "      epochs = 20,\n",
        "      callbacks = list(my_cb),\n",
        "      verbose = 0)\n",
        "\n",
        "plot(my_history)"
      ],
      "id": "a2dc5dbf-1555-4d80-966f-bb8b483552ea"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model %>%\n",
        "  evaluate(x = x_test2d, y = y_test)"
      ],
      "id": "cea324eb-6676-48f4-873c-01842a612258"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- keras_model_sequential() %>%\n",
        "  layer_conv_2d(filters = 20, kernel_size = 5, activation = \"relu\",\n",
        "                input_shape = c(28, 28, 1)) %>%\n",
        "  layer_max_pooling_2d(pool_size = 2, strides = 2) %>%\n",
        "  layer_conv_2d(filters = 50, kernel_size = 5, activation = \"relu\") %>%\n",
        "  layer_max_pooling_2d(pool_size = 2, strides = 2) %>%\n",
        "  layer_dropout(rate = 0.25) %>%\n",
        "  layer_flatten() %>%\n",
        "  layer_dense(units = 500, activation = \"relu\") %>%\n",
        "  layer_dropout(rate = 0.5) %>%\n",
        "  layer_dense(units = 10, activation = \"softmax\")\n",
        "\n",
        "my_model %>% compile(\n",
        "  loss = \"sparse_categorical_crossentropy\",\n",
        "  optimizer = \"rmsprop\",\n",
        "  metrics = c(\"accuracy\"))\n",
        "\n",
        "my_cb <- callback_early_stopping(patience = 5,\n",
        "                                 restore_best_weights = TRUE)"
      ],
      "id": "61beec17-8ee2-4d4a-9555-d0294e10c79d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_history <- my_model %>%\n",
        "  fit(x = x_train2d,\n",
        "      y = y_train,\n",
        "      validation_split = 0.2,\n",
        "      batch_size = 128,\n",
        "      epochs = 20,\n",
        "      callbacks = list(my_cb),\n",
        "      verbose = 0)\n",
        "\n",
        "plot(my_history)"
      ],
      "id": "e2842581-fcda-48c8-b500-740ecec489eb"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model %>%\n",
        "  evaluate(x = x_test2d, y = y_test)"
      ],
      "id": "34e5e114-e03c-4096-8e34-7b3fc3fe553c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y_prob <- my_model %>% predict(x_test2d) # \u30ab\u30c6\u30b4\u30ea\u306b\u5c5e\u3059\u308b\u78ba\u7387\n",
        "\n",
        "my_result <- data.frame(\n",
        "  y_prob = apply(y_prob, 1, max),           # \u78ba\u7387\u306e\u6700\u5927\u5024\n",
        "  y_     = apply(y_prob, 1, which.max) - 1, # \u4e88\u6e2c\u30ab\u30c6\u30b4\u30ea\n",
        "  y      = y_test,                          # \u6b63\u89e3\n",
        "  id     = seq_len(length(y_test))) %>%     # \u756a\u53f7\n",
        "  filter(y_ != y) %>%                       # \u4e88\u6e2c\u304c\u306f\u305a\u308c\u305f\u3082\u306e\u3092\u6b8b\u3059\n",
        "  arrange(desc(y_prob))                     # \u78ba\u7387\u306e\u5927\u304d\u3044\u9806\u306b\u4e26\u3073\u66ff\u3048\u308b"
      ],
      "id": "a1b64129-1b4f-4330-9076-efbbbfccf7c0"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "head(my_result)"
      ],
      "id": "95b3e861-f5c3-4130-9aff-d49f4a85e2ff"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- my_result[1:5, ]$id\n",
        "my_labels <- sprintf(\"%s (%s)\",\n",
        "  my_result[1:5, ]$y, tmp)\n",
        "my_fig <- expand.grid(\n",
        "  label = my_labels,\n",
        "  y = 28:1,\n",
        "  x = 1:28)\n",
        "my_fig$z <- as.vector(\n",
        "  x_test[tmp, , ])\n",
        "\n",
        "my_fig %>% ggplot(\n",
        "  aes(x = x, y = y, fill = z)) +\n",
        "  geom_raster() +\n",
        "  coord_fixed() +\n",
        "  theme_void() +\n",
        "  theme(legend.position = \"none\") +\n",
        "  facet_grid(. ~ label)"
      ],
      "id": "6024feea-7ee2-44c9-aa70-2817f6eea133"
    },
    {
      "id": "8dddef7b",
      "cell_type": "markdown",
      "source": "## 11.4 AutoML",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(h2o)\n",
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  library(keras3)\n",
        "} else {\n",
        "  library(keras)\n",
        "}\n",
        "library(tidyverse)\n",
        "\n",
        "h2o.init()\n",
        "h2o.no_progress()\n",
        "# h2o.shutdown(prompt = FALSE) # \u505c\u6b62"
      ],
      "id": "15559443-5ef4-4fdf-bdfc-8e614666f495"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_url <- str_c(\"https://raw.githubusercontent.com/taroyabuki\",\n",
        "                \"/fromzero/master/data/wine.csv\")\n",
        "my_data <- read_csv(my_url)\n",
        "my_frame <- as.h2o(my_data) # \u901a\u5e38\u306e\u30c7\u30fc\u30bf\u30d5\u30ec\u30fc\u30e0\u3092H2OFrame\u306b\u5909\u63db\u3059\u308b\uff0e\n",
        "# \u3042\u308b\u3044\u306f\n",
        "my_frame <- h2o.importFile(my_url, header = TRUE) # \u30c7\u30fc\u30bf\u3092\u8aad\u307f\u8fbc\u3080\uff0e"
      ],
      "id": "a575cee8-8c66-48a4-b41b-76d66067d525"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_frame\n",
        "\n",
        "# \u901a\u5e38\u306e\u30c7\u30fc\u30bf\u30d5\u30ec\u30fc\u30e0\u306b\u623b\u3059\uff0e\n",
        "my_frame %>% as.data.frame %>% head\n",
        "# \u7d50\u679c\u306f\u5272\u611b\uff08\u898b\u305f\u76ee\u306f\u540c\u3058\uff09"
      ],
      "id": "ba2aa12e-4398-4e09-ad16-5e1e7a5a752c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- h2o.automl(\n",
        "    y = \"LPRICE2\",\n",
        "    training_frame = my_frame,\n",
        "    max_runtime_secs = 60)"
      ],
      "id": "78ba63a7-84d9-4532-aa8e-44d5864321fa"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "min(my_model@leaderboard$rmse)"
      ],
      "id": "e2033b89-9df3-4cae-ba64-6bc5f6ecd55e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- my_model %>%\n",
        "  predict(my_frame) %>%\n",
        "  as.data.frame\n",
        "y_ <- tmp$predict\n",
        "y  <- my_data$LPRICE2\n",
        "\n",
        "plot(y, y_)"
      ],
      "id": "1746da09-4ccf-402e-a8be-dd7a9c7a831b"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()\n",
        "my_index <- sample(1:60000, 6000)\n",
        "x_train <- x_train[my_index, , ]\n",
        "y_train <- y_train[my_index]"
      ],
      "id": "9bd4ca7c-e98f-41d7-a52b-7979c260ced2"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- x_train %>%\n",
        "  array_reshape(c(-1, 28 * 28)) %>%\n",
        "  as.data.frame\n",
        "tmp$y <- as.factor(y_train)\n",
        "my_train <- as.h2o(tmp)\n",
        "\n",
        "tmp <- x_test %>%\n",
        "  array_reshape(c(-1, 28 * 28)) %>%\n",
        "  as.data.frame\n",
        "my_test <- as.h2o(tmp)"
      ],
      "id": "e2d5b58c-672e-4ea0-936c-8a2946488d0f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_model <- h2o.automl(\n",
        "    y = \"y\",\n",
        "    training_frame = my_train,\n",
        "    max_runtime_secs = 120)"
      ],
      "id": "954bd63a-074f-4ec8-afca-dcf1f90bf520"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "min(my_model@leaderboard$\n",
        "    mean_per_class_error)"
      ],
      "id": "dc59347a-9c2d-46e9-b063-d0c697fa8ac5"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- my_model %>%\n",
        "  predict(my_test) %>% as.data.frame\n",
        "y_ <- tmp$predict\n",
        "\n",
        "mean(y_ == y_test)"
      ],
      "id": "1e10a537-b15d-4ef7-9a90-bea67baa6d82"
    },
    {
      "id": "068db036",
      "cell_type": "markdown",
      "source": "# 12 \u6642\u7cfb\u5217\u4e88\u6e2c",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Google Colaboratory\u306e\u74b0\u5883\u8a2d\u5b9a\n",
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  options(Ncpus = parallel::detectCores())\n",
        "  installed_packages <- rownames(installed.packages())\n",
        "  packages_to_install <- c(\"caret\", \"fable\", \"feasts\", \"prophet\", \"tsibble\", \"urca\")\n",
        "  install.packages(setdiff(packages_to_install, installed_packages))\n",
        "  install.packages(c(\"ggplot2\"))\n",
        "}"
      ],
      "id": "54837683-a705-4e4e-8979-141d3be7849f"
    },
    {
      "id": "e008c70e",
      "cell_type": "markdown",
      "source": "## 12.1 \u65e5\u6642\u3068\u65e5\u6642\u306e\u5217",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "as.POSIXct(\"2021-01-01\")"
      ],
      "id": "ad9f3844-56e5-4048-936f-e5f7e8a1fbe7"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tsibble)\n",
        "\n",
        "seq(from = 2021, to = 2023, by = 1)\n",
        "\n",
        "seq(from = yearmonth(\"202101\"), to = yearmonth(\"202103\"), by = 2)\n",
        "\n",
        "seq(from = as.POSIXct(\"2021-01-01\"), to = as.POSIXct(\"2021-01-03\"), by = \"1 day\")\n",
        "\n",
        "seq(from = as.POSIXct(\"2021-01-01 00:00:00\"),\n",
        "    to   = as.POSIXct(\"2021-01-01 03:00:00\"), by = \"2 hour\")"
      ],
      "id": "ff3f39e6-75b9-40a8-ace7-fdcd01c35a7f"
    },
    {
      "id": "d7f4ac91",
      "cell_type": "markdown",
      "source": "## 12.2 \u6642\u7cfb\u5217\u30c7\u30fc\u30bf\u306e\u4e88\u6e2c",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_data <- as.vector(AirPassengers)"
      ],
      "id": "27e492cf-294d-48a2-be39-a1be70d9042a"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "n <- length(my_data) # \u30c7\u30fc\u30bf\u6570\uff08144\uff09\n",
        "k <- 108             # \u8a13\u7df4\u30c7\u30fc\u30bf\u6570"
      ],
      "id": "861b0c12-c582-47e5-8648-fa9eda4c9149"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "library(tsibble)\n",
        "\n",
        "my_ds <- seq(\n",
        "  from = yearmonth(\"1949/01\"),\n",
        "  to   = yearmonth(\"1960/12\"),\n",
        "  by   = 1)\n",
        "my_label <- rep(\n",
        "  c(\"train\", \"test\"),\n",
        "  c(k, n - k))\n",
        "my_df <- tsibble(\n",
        "  ds    = my_ds,\n",
        "  x     = 0:(n - 1),\n",
        "  y     = my_data,\n",
        "  label = my_label,\n",
        "  index = ds) # \u65e5\u6642\u306e\u5217\u306e\u6307\u5b9a\n",
        "\n",
        "head(my_df)"
      ],
      "id": "d9df9740-913f-41b5-9486-6c25b62af78f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_train <- my_df[  1:k , ]\n",
        "my_test  <- my_df[-(1:k), ]\n",
        "y <- my_test$y"
      ],
      "id": "4c37533e-54b3-4797-bc08-9117a04cb749"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_plot <- my_df %>%\n",
        "  ggplot(aes(x = ds,\n",
        "             y = y,\n",
        "             color = label)) +\n",
        "  geom_line()\n",
        "my_plot"
      ],
      "id": "5c6e2c9a-f6fb-45e1-9418-7afcfdb0940f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(caret)\n",
        "my_lm_model <- train(form = y ~ x, data = my_train, method = \"lm\")\n",
        "y_ <- my_lm_model %>% predict(my_test)\n",
        "caret::RMSE(y, y_) # RMSE\uff08\u30c6\u30b9\u30c8\uff09"
      ],
      "id": "4e679949-5317-44dc-9582-25862f766c9f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y_ <- my_lm_model %>% predict(my_df)\n",
        "tmp <- my_df %>%\n",
        "  mutate(y = y_, label = \"model\")\n",
        "my_plot + geom_line(data = tmp)"
      ],
      "id": "bfdeef72-f5a5-4553-83bf-ffff5ec30560"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(fable)\n",
        "my_arima_model <- my_train %>% model(ARIMA(y))\n",
        "my_arima_model"
      ],
      "id": "0794790e-c3cb-4482-b59b-e87fde83f762"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- my_arima_model %>% forecast(h = \"3 years\")\n",
        "head(tmp)"
      ],
      "id": "d5d1163f-dd1e-497d-b65f-95194fcf3a09"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y_ <- tmp$.mean\n",
        "caret::RMSE(y_, y)"
      ],
      "id": "56045aac-fec7-45ae-ac80-b4a1bb21927d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# \u4e88\u6e2c\u7d50\u679c\u306e\u307f\u3067\u3088\u3044\u5834\u5408\n",
        "#tmp %>% autoplot\n",
        "\n",
        "tmp %>% autoplot +\n",
        "  geom_line(data = my_df,\n",
        "            aes(x = ds,\n",
        "                y = y,\n",
        "                color = label))"
      ],
      "id": "1a58982c-eda6-418f-b758-b53ef946ffa2"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(prophet)\n",
        "my_prophet_model <- my_train %>%\n",
        "  prophet(seasonality.mode = \"multiplicative\")"
      ],
      "id": "d12ca9ee-0e50-42d1-bcfb-feb7572d153d"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tmp <- my_prophet_model %>% predict(my_test)\n",
        "head(tmp[, c(\"ds\", \"yhat\", \"yhat_lower\", \"yhat_upper\")])"
      ],
      "id": "f60d6e11-5c43-4d7c-9ea1-b21b1714c14c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "y_ <- tmp$yhat\n",
        "caret::RMSE(y_, y)"
      ],
      "id": "7d8f60d2-2d87-4647-ae12-054226c49ac1"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# my_prophet_model %>% plot(tmp) # \u4e88\u6e2c\u7d50\u679c\u306e\u307f\u3067\u3088\u3044\u5834\u5408\n",
        "\n",
        "my_prophet_model %>% plot(tmp) +\n",
        "  geom_line(data = my_train, aes(x = as.POSIXct(ds))) +\n",
        "  geom_line(data = my_test,  aes(x = as.POSIXct(ds)), color = \"red\")"
      ],
      "id": "fabd0ff5-0b66-4041-83ed-dd7a7c91e289"
    },
    {
      "id": "e0932d28",
      "cell_type": "markdown",
      "source": "# 13 \u6559\u5e2b\u306a\u3057\u5b66\u7fd2",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Google Colaboratory\u306e\u74b0\u5883\u8a2d\u5b9a\n",
        "if (Sys.getenv(\"COLAB_JUPYTER_IP\") != \"\") {\n",
        "  options(Ncpus = parallel::detectCores())\n",
        "  installed_packages <- rownames(installed.packages())\n",
        "  packages_to_install <- c(\"factoextra\", \"ggbiplot\", \"igraph\")\n",
        "  install.packages(setdiff(packages_to_install, installed_packages))\n",
        "}"
      ],
      "id": "4020e718-150f-4a08-9244-6145a7c8e877"
    },
    {
      "id": "b7d4b5b5",
      "cell_type": "markdown",
      "source": "## 13.1 \u4e3b\u6210\u5206\u5206\u6790",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "\n",
        "my_data <- data.frame(\n",
        "  language  = c(  0,  20,  20,  25,  22,  17),\n",
        "  english   = c(  0,  20,  40,  20,  24,  18),\n",
        "  math      = c(100,  20,   5,  30,  17,  25),\n",
        "  science   = c(  0,  20,   5,  25,  16,  23),\n",
        "  society   = c(  0,  20,  30,   0,  21,  17),\n",
        "  row.names = c(\"A\", \"B\", \"C\", \"D\", \"E\", \"F\"))\n",
        "my_result <- my_data %>% prcomp # \u4e3b\u6210\u5206\u5206\u6790\u306e\u5b9f\u884c"
      ],
      "id": "1b1f2cd2-09f2-423f-9412-a10530dacfb0"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_result$x # \u4e3b\u6210\u5206\u30b9\u30b3\u30a2"
      ],
      "id": "c0572ffa-5b7a-41ed-8404-a094d12cadc4"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_result %>% ggbiplot::ggbiplot(\n",
        "  labels = row.names(my_data),\n",
        "  scale = 0)"
      ],
      "id": "917f1f11-26c2-417a-80d5-6bcdf1e89d2f"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_result$rotation %>% t"
      ],
      "id": "a30b76c1-8b65-4fd8-b16e-05517e27ead6"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "summary(my_result)"
      ],
      "id": "6623042c-7014-46c3-8fd0-27aebc9c8ac9"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_result <- prcomp(\n",
        "  x = my_data,\n",
        "  scale = TRUE)       # \u6a19\u6e96\u5316\n",
        "# \u3042\u308b\u3044\u306f\n",
        "my_result <- prcomp(\n",
        "  x = scale(my_data)) # \u6a19\u6e96\u5316\u30c7\u30fc\u30bf\n",
        "\n",
        "my_result$x # \u4e3b\u6210\u5206\u30b9\u30b3\u30a2"
      ],
      "id": "695fd567-3535-49e4-b247-35e1de63e114"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "Z  <- my_data %>% scale(scale = FALSE) %>% as.matrix # \u6a19\u6e96\u5316\u3057\u306a\u3044\u5834\u5408\n",
        "#Z <- my_data %>% scale(scale = TRUE)  %>% as.matrix # \u6a19\u6e96\u5316\u3059\u308b\u5834\u5408\n",
        "\n",
        "n <- nrow(my_data)\n",
        "S <- var(Z)                          # \u5206\u6563\u5171\u5206\u6563\u884c\u5217\n",
        "#S <- t(Z) %*% Z / (n - 1)           # \uff08\u540c\u3058\u7d50\u679c\uff09\n",
        "tmp <- eigen(S)                      # \u56fa\u6709\u5024\u3068\u56fa\u6709\u30d9\u30af\u30c8\u30eb\n",
        "Z %*% tmp$vectors                    # \u4e3b\u6210\u5206\u30b9\u30b3\u30a2\uff08\u7d50\u679c\u306f\u5272\u611b\uff09\n",
        "cumsum(tmp$values) / sum(tmp$values) # \u7d2f\u7a4d\u5bc4\u4e0e\u7387"
      ],
      "id": "a9ce5e66-e3a5-431a-9240-6e2de900acce"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "udv <- svd(Z) # \u7279\u7570\u5024\u5206\u89e3\n",
        "U <- udv$u\n",
        "d <- udv$d\n",
        "V <- udv$v\n",
        "W <- diag(d)\n",
        "\n",
        "c(all.equal(Z, U %*% W %*% t(V), check.attributes = FALSE), # \u78ba\u8a8d1\n",
        "  all.equal(t(U) %*% U, diag(dim(U)[2])),                   # \u78ba\u8a8d2\n",
        "  all.equal(t(V) %*% V, diag(dim(V)[2])))                   # \u78ba\u8a8d3\n",
        "\n",
        "U %*% W            # \u4e3b\u6210\u5206\u30b9\u30b3\u30a2\uff08\u7d50\u679c\u306f\u5272\u611b\uff09\n",
        "\n",
        "e <- d^2 / (n - 1) # \u5206\u6563\u5171\u5206\u6563\u884c\u5217\u306e\u56fa\u6709\u5024\n",
        "cumsum(e) / sum(e) # \u7d2f\u7a4d\u5bc4\u4e0e\u7387"
      ],
      "id": "1dd46ed1-9e20-434a-8b8f-98a708aaa56a"
    },
    {
      "id": "ac0acb3d",
      "cell_type": "markdown",
      "source": "## 13.2 \u30af\u30e9\u30b9\u30bf\u5206\u6790",
      "metadata": {}
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "\n",
        "my_data <- data.frame(\n",
        "  x         = c(  0, -16,  10,  10),\n",
        "  y         = c(  0,   0,  10, -15),\n",
        "  row.names = c(\"A\", \"B\", \"C\", \"D\"))\n",
        "\n",
        "my_result <- my_data %>%\n",
        "  dist(\"euclidian\") %>% # dist\u3060\u3051\u3067\u3082\u53ef\n",
        "  hclust(\"complete\")    # hclust\u3060\u3051\u3067\u3082\u53ef"
      ],
      "id": "692ff442-e3dd-4138-a2a3-0b004de2a720"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_result %>% factoextra::fviz_dend(\n",
        "  k = 3, # \u30af\u30e9\u30b9\u30bf\u6570\n",
        "  rect = TRUE, rect_fill = TRUE)"
      ],
      "id": "f7ea8927-f600-4d54-a584-8da84a31fa17"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_result %>% factoextra::fviz_dend(\n",
        "  k = 3,\n",
        "  rect = TRUE, rect_fill = TRUE,\n",
        "  type = \"phylogenic\")"
      ],
      "id": "8f7abe2c-c29f-4ccd-b7ba-5ea1f6d214b3"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_result %>% cutree(3)"
      ],
      "id": "153436e1-ce1e-4fa2-b7e9-1f9310d50675"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "\n",
        "my_data <- data.frame(\n",
        "  language  = c(  0,  20,  20,  25,  22,  17),\n",
        "  english   = c(  0,  20,  40,  20,  24,  18),\n",
        "  math      = c(100,  20,   5,  30,  17,  25),\n",
        "  science   = c(  0,  20,   5,  25,  16,  23),\n",
        "  society   = c(  0,  20,  30,   0,  21,  17),\n",
        "  row.names = c(\"A\", \"B\", \"C\", \"D\", \"E\", \"F\"))\n",
        "\n",
        "try( # RMarkdown\u3067\u767a\u751f\u3059\u308b\u30a8\u30e9\u30fc\u3092\u56de\u907f\u3059\u308b\uff0e\n",
        "  my_data %>% scale %>%                        # \u5217\u3054\u3068\u306e\u6a19\u6e96\u5316\n",
        "    gplots::heatmap.2(cexRow = 1, cexCol = 1), # \u30e9\u30d9\u30eb\u306e\u30b5\u30a4\u30ba\u3092\u6307\u5b9a\u3057\u3066\u63cf\u753b\u3059\u308b\uff0e\n",
        "  silent = TRUE)"
      ],
      "id": "92c90661-0686-4815-99bf-5f123efd8b6e"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "\n",
        "my_data <- data.frame(\n",
        "  x         = c(  0, -16,  10,  10),\n",
        "  y         = c(  0,   0,  10, -15),\n",
        "  row.names = c(\"A\", \"B\", \"C\", \"D\"))\n",
        "\n",
        "my_result <- my_data %>% kmeans(3)"
      ],
      "id": "c07e549e-b30f-44ee-b9b9-9c72ab3b05bf"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "my_result$cluster"
      ],
      "id": "81777fcc-053b-4e3a-b80f-e99fe6589435"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "library(factoextra)\n",
        "\n",
        "my_data <- iris[, -5]\n",
        "\n",
        "f <- 2:5 %>% map(function(k) {\n",
        "  my_data %>% kmeans(k) %>%\n",
        "    fviz_cluster(data = my_data, geom = \"point\") +\n",
        "    ggtitle(sprintf(\"k = %s\", k))\n",
        "})\n",
        "gridExtra::grid.arrange(f[[1]], f[[2]], f[[3]], f[[4]], ncol = 2)"
      ],
      "id": "1fe1ecc7-7022-4f92-bb6c-9a3a7be221b2"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "fviz_nbclust(my_data, kmeans, method = \"wss\")"
      ],
      "id": "a972e379-5102-4b81-a2c9-bc1c5d986f4c"
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "library(tidyverse)\n",
        "my_data <- iris[, -5] %>% scale\n",
        "\n",
        "my_result <- prcomp(my_data)$x %>% as.data.frame # \u4e3b\u6210\u5206\u5206\u6790\n",
        "\n",
        "# \u975e\u968e\u5c64\u7684\u30af\u30e9\u30b9\u30bf\u5206\u6790\u306e\u5834\u5408\n",
        "my_result$cluster <- (my_data %>% scale %>% kmeans(3))$cluster %>% as.factor\n",
        "\n",
        "# \u968e\u5c64\u7684\u30af\u30e9\u30b9\u30bf\u5206\u6790\u306e\u5834\u5408\n",
        "#my_result$cluster <- my_data %>% dist %>% hclust %>% cutree(3) %>% as.factor\n",
        "\n",
        "my_result %>%\n",
        "  ggplot(aes(x = PC1, y = PC2, color = cluster)) + # \u8272\u3067\u30af\u30e9\u30b9\u30bf\u3092\u8868\u73fe\u3059\u308b\uff0e\n",
        "  geom_point(shape = iris$Species) +               # \u5f62\u3067\u54c1\u7a2e\u3092\u8868\u73fe\u3059\u308b\uff0e\n",
        "  theme(legend.position = \"none\")"
      ],
      "id": "def687bc-5952-4bf1-9506-956639957af1"
    }
  ],
  "nbformat": 4,
  "nbformat_minor": 5,
  "metadata": {
    "kernelspec": {
      "name": "ir",
      "display_name": "R"
    }
  }
}