rate_purchase_per_user.py

""" Purchase rate per user, normalized by when they first joined the network.
"""

import altair as alt
import datetime
import pandas as pd

import infra.constants
import infra.dask
import infra.pd


def make_rate_chart():
    transactions = infra.pd.read_parquet("data/clean/transactions_TM.parquet")

    # Each user's total amount of data purchased directly.
    purchases = transactions.loc[transactions["kind"] == "purchase"]

    # Find the first day the user was active. Define "active" as making first
    # purchase or first data in network.
    user_active_ranges = infra.pd.read_parquet("data/clean/user_active_deltas.parquet")[["user", "days_since_first_active", "days_active"]]

    # Drop users that have been active less than a week.
    users_to_analyze = user_active_ranges.loc[
        user_active_ranges["days_since_first_active"] >= 7,
        ]

    # Drop users active for less than one week
    users_to_analyze = users_to_analyze.loc[
        users_to_analyze["days_active"] >=1,
    ]

    aggregated_purchases = purchases.groupby(
        "user"
    )[["amount_idr", "amount_bytes"]].sum().reset_index()
    print(len(aggregated_purchases))

    aggregated_purchases["amount_GB"] = aggregated_purchases["amount_bytes"] * float(1) / (1000 ** 3)
    aggregated_purchases["amount_USD"] = aggregated_purchases["amount_idr"] * infra.constants.IDR_TO_USD

    # Merge in the active times
    aggregated_purchases = aggregated_purchases.merge(users_to_analyze,
                                                      on="user",
                                                      how="inner")
    # Compute USD/Day
    aggregated_purchases["USD_per_day"] = (
            aggregated_purchases["amount_USD"] /
            aggregated_purchases["days_active"]
    )

    # # Drop un-needed columns since altair cannot handle timedelta types.
    # aggregated_purchases = aggregated_purchases[["user", "USD_per_day", "days_active"]]
    print(aggregated_purchases)

    alt.Chart(aggregated_purchases).mark_circle(opacity=0.7).encode(
        x=alt.X('days_active:Q',
                sort=alt.SortField(field="amount_USD",
                                   order="descending"
                                   ),
                axis=alt.Axis(labels=True),
                title="Days Active"
                ),
        y=alt.Y('USD_per_day',
                title= "Average Daily Purchase (USD)"
                #scale=alt.Scale(type="log"),
                ),
        color=alt.Color(
            "amount_USD:Q",
            title="Total (USD)",
            scale=alt.Scale(scheme="viridis"),
        ),
        size=alt.Size(
            "amount_USD:Q",
            title="Total (USD)",
            scale=alt.Scale(domain=[0, 2000], range=[30, 1000]),
            legend=None,
        ),
    ).properties(
        width=600,
    ).save("renders/rate_purchase_per_user.png", scale_factor=2.0)

    # Compute a CDF since the specific user does not matter
    value_column = "USD_per_day"

    # Find the PDF first
    stats_frame = aggregated_purchases.groupby(value_column).count()[["user"]].rename(columns = {"user": "user_count"})
    stats_frame["pdf"] = stats_frame["user_count"] / sum(stats_frame["user_count"])
    stats_frame["cdf"] = stats_frame["pdf"].cumsum()

    stats_frame = stats_frame.reset_index()
    alt.Chart(stats_frame).mark_line(interpolate="step-after", clip=True).encode(
        x=alt.X('USD_per_day:Q',
                scale=alt.Scale(type="linear"),
                title="Amount Purchased/Days Active (USD/Day)",
                ),
        y=alt.Y('cdf',
                title="CDF of Users N={}".format(len(aggregated_purchases)),
                scale=alt.Scale(domain=[0, 1]),
                ),
    ).properties(
        width=500,
        height=200,
    ).save("renders/rate_purchase_per_user_cdf.png", scale_factor=2.0)

    alt.Chart(stats_frame).mark_line(interpolate="step-after", clip=True).encode(
        x=alt.X('USD_per_day:Q',
                scale=alt.Scale(type="log"),
                title="Amount Purchased/Days Active (USD/Day - Log Scale)",
                ),
        y=alt.Y('cdf',
                title="CDF of Users N={}".format(len(aggregated_purchases)),
                scale=alt.Scale(domain=[0, 1]),
                ),
    ).properties(
        width=500,
        height=200,
    ).save("renders/rate_purchase_per_user_cdf_log.png", scale_factor=2.0)


if __name__ == "__main__":
    # Module specific format options
    pd.set_option('display.max_columns', None)
    pd.set_option('display.width', None)
    pd.set_option('display.max_rows', None)
    make_rate_chart()