example/Time_Series_Classification_and_Explanation_with_MrSQM.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "auburn-vienna",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-12T14:59:17.906501Z",
     "start_time": "2023-05-12T14:59:17.898789Z"
    }
   },
   "outputs": [],
   "source": [
    "from mrsqm import MrSQMClassifier, MrSQMTransformer\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn import metrics\n",
    "from scipy.interpolate import interp1d\n",
    "import matplotlib.pyplot as plt\n",
    "import timeit\n",
    "\n",
    "from sklearn.linear_model import RidgeClassifierCV\n",
    "\n",
    "from aeon.datasets import load_classification  "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "affected-schema",
   "metadata": {},
   "source": [
    "## Load dataset\n",
    "\n",
    "Load dataset with *aeon*:\n",
    "https://www.aeon-toolkit.org/en/latest/api_reference/auto_generated/aeon.datasets.load_classification.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "announced-demand",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-12T14:49:42.255558Z",
     "start_time": "2023-05-12T14:49:42.240222Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Coffee\n",
      "(28, 1, 286) (28,)\n",
      "(28, 1, 286) (28,)\n"
     ]
    }
   ],
   "source": [
    "ds = \"Coffee\"\n",
    "X_train, y_train = load_classification(name=ds,split=\"train\")\n",
    "X_test, y_test = load_classification(name=ds,split=\"test\")\n",
    "\n",
    "print(ds)\n",
    "print(X_train.shape, y_train.shape)\n",
    "print(X_test.shape, y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "right-tracy",
   "metadata": {},
   "source": [
    "## Train with MrSQMClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "contained-theorem",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-12T14:50:09.688632Z",
     "start_time": "2023-05-12T14:50:09.469954Z"
    }
   },
   "outputs": [],
   "source": [
    "clf = MrSQMClassifier().fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "willing-integer",
   "metadata": {},
   "source": [
    "## Test the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "split-species",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-12T14:50:11.030256Z",
     "start_time": "2023-05-12T14:50:10.965630Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9642857142857143\n"
     ]
    }
   ],
   "source": [
    "y_pred = clf.predict(X_test)\n",
    "print(metrics.accuracy_score(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "527995f5",
   "metadata": {},
   "source": [
    "## Use MrSQM Transformer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "8172fd38",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-12T14:52:05.971574Z",
     "start_time": "2023-05-12T14:52:05.966834Z"
    }
   },
   "outputs": [],
   "source": [
    "tfm = MrSQMTransformer()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "8fe4ee44",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-12T14:52:06.734512Z",
     "start_time": "2023-05-12T14:52:06.379881Z"
    }
   },
   "outputs": [],
   "source": [
    "X_train_transform = tfm.fit_transform(X_train,y_train)\n",
    "X_test_transform = tfm.transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "495247bc",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-12T15:03:33.986523Z",
     "start_time": "2023-05-12T15:03:33.968424Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9642857142857143\n"
     ]
    }
   ],
   "source": [
    "# use ridgecv classifier\n",
    "ridge = RidgeClassifierCV(alphas = np.logspace(-3, 3, 10)).fit(X_train_transform,y_train)\n",
    "y_pred = ridge.predict(X_test_transform)\n",
    "print(metrics.accuracy_score(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "statistical-building",
   "metadata": {},
   "source": [
    "## Explanation with saliency map\n",
    "A saliency map is basically a weight vector for each point in the time series corresponding to a class. Higher weight means the data point is more relevant to the class according to the model.\n",
    "\n",
    "*get_saliency_map* return a c X L array where c is the number of class and L is the length of the time series.\n",
    "\n",
    "It is important to note that in MrSQM saliency map only works if there is at least 1 SAX representation used for training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "lyric-vinyl",
   "metadata": {},
   "outputs": [],
   "source": [
    "# This example visualise the saliency map of the 1st sample (class '0') in the Coffee dataset \n",
    "clf = MrSQMClassifier(nsax=5,nsfa=0).fit(X_train,y_train)\n",
    "x = X_train[0,0,:]\n",
    "w = clf.get_saliency_map(x)[0,:]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from scipy.interpolate import interp1d\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "\n",
    "def transform(X):\n",
    "    ma,mi = np.max(X), np.min(X)\n",
    "    X = (X - mi)/(ma-mi)\n",
    "    return X*100\n",
    "weight = abs(w)\n",
    "weight = transform(weight)\n",
    "ts = np.squeeze(x)\n",
    "   \n",
    "max_length1, max_length2 = ts.shape[0],10000 #\n",
    "x1 = np.linspace(0,max_length1,num = max_length1)\n",
    "x2 = np.linspace(0,max_length1,num = max_length2)\n",
    "y1 = ts\n",
    "\n",
    "f = interp1d(x1, y1) # interpolate time series\n",
    "fcas = interp1d(x1, weight) # interpolate weight color\n",
    "weight = fcas(x2) # convert vector of original weight vector to new weight vector\n",
    "\n",
    "plt.scatter(x2,f(x2), c = weight, cmap = 'jet', marker='.', s= 1,vmin=0,vmax = 100)\n",
    "# plt.xlabel('Explanation for index %d, dataset %s' %(idx, ds))\n",
    "cbar = plt.colorbar(orientation = 'vertical')\n",
    "\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "crazy-brunei",
   "metadata": {},
   "source": [
    "## Control the number of SAX and SFA representations\n",
    "The parameters *nsax* and *nsfa* control the number of SAX and SFA representations produced from the raw data. The number of representations is *nsax* (or *nsfa*) x log(L) where L is the length of the time series.\n",
    "\n",
    "Adding more representations (bigger *nsax* or *nsfa*) can improve the accuracy but slow down the training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "empty-violation",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9642857142857143\n",
      "Time to train and predict (sec): 1.340001734904945\n"
     ]
    }
   ],
   "source": [
    "starttime = timeit.default_timer()\n",
    "clf = MrSQMClassifier(nsax=1, nsfa=3).fit(X_train,y_train)\n",
    "y_pred = clf.predict(X_test)\n",
    "print(metrics.accuracy_score(y_test, y_pred))\n",
    "print(\"Time to train and predict (sec):\", timeit.default_timer() - starttime)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "moving-tennessee",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.0\n",
      "Time to train and predict (sec): 0.29645546339452267\n"
     ]
    }
   ],
   "source": [
    "#parameters about training that the user can change\n",
    "'''     \n",
    "    Overview: MrSQM is an efficient time series classifier utilizing symbolic representations of time series. MrSQM implements four different feature selection strategies (R,S,RS,SR) that can quickly select subsequences from multiple symbolic representations of time series data.\n",
    "    def __init__(self, strat = 'RS', features_per_rep = 500, selection_per_rep = 2000, nsax = 1, nsfa = 0, custom_config=None, random_state = None, sfa_norm = True):\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    \n",
    "    strat               : str, feature selection strategy, either 'R','S','SR', or 'RS'. R and S are single-stage filters while RS and SR are two-stage filters. By default set to 'RS'.\n",
    "    features_per_rep    : int, (maximum) number of features selected per representation. By deafault set to 500.\n",
    "    selection_per_rep   : int, (maximum) number of candidate features selected per representation. Only applied in two stages strategies (RS and SR). By deafault set to 2000.\n",
    "    nsax                : int, control the number of representations produced by sax transformation.\n",
    "    nsfa                : int, control the number of representations produced by sfa transformation.\n",
    "    custom_config       : dict, customized parameters for the symbolic transformation.\n",
    "    random_state        : set random seed for classifier. By default 'none'.\n",
    "    ts_norm             : time series normalisation (standardisation). By default set to 'True'.\n",
    "    '''\n",
    "\n",
    "starttime = timeit.default_timer()\n",
    "clf = MrSQMClassifier(random_state=42,nsax=0, nsfa=1, sfa_norm = 'False').fit(X_train,y_train)\n",
    "y_pred = clf.predict(X_test)\n",
    "print(metrics.accuracy_score(y_test, y_pred))\n",
    "print(\"Time to train and predict (sec):\", timeit.default_timer() - starttime)\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}