strava_cummax.py

# Export data using https://www.strava.com/athlete/delete_your_account
# Yes, the data export option is on the Delete Your Account page.
#
# Run this script on the 'activities.csv'.
import sys

from datetime import datetime, timedelta

import matplotlib.pyplot as plt
import numpy as np
import pandas


DATE_FMT = r"%b %d, %Y, %I:%M:%S %p"


def _parse_date(val: str) -> datetime:
    return datetime.strptime(val, DATE_FMT)


def _convert_date(col) -> np.ndarray:
    return np.array([_convert_date(value) for value in col])


CONVERTERS = {
    "Activity Date": _parse_date,
}


def load(filepath: str) -> pandas.DataFrame:
    with open(filepath) as datafile:
        return pandas.read_csv(datafile, converters=CONVERTERS)


def describe(row):
    time = timedelta(seconds=int(row["Elapsed Time"]))
    dist = row["Distance"]
    speed = row["Average Speed"] * 3600 / 1000  # m/s to km/h
    name = row["Activity Name"]
    date = row["Activity Date"]
    aid = row["Activity ID"]

    print(f"{date} {name}:\n\t{dist} km   {speed:.2f} km/h   {time}\n\thttps://strava.com/activities/{aid}\n")


def plot_distance(data: pandas.DataFrame):
    plt.figure()
    ax = plt.axes()

    dates = data["Activity Date"]
    dist = data["Distance"]
    maxen = dist.cummax()

    ax.bar(dates, dist, label="activity distance")
    ax.plot(dates, maxen, "--", label="cumulative maximum")

    # find indexes that changed max
    changes = np.flatnonzero(np.diff(maxen)) + 1

    print("Running maximum distance")
    for idx in changes:
        describe(data.iloc[idx])
        plt.text(dates[idx], maxen[idx]+0.5, str(maxen[idx]), size=16, ha="center")

    plt.ylabel("Distance (km)")
    plt.xlabel("Date")

    plt.legend()
    plt.show()


def main():
    plt.style.use('tableau-colorblind10')
    fname = sys.argv[1]
    print(f"Loading data from {fname}")
    data = load(fname)
    plot_distance(data)


if __name__ == "__main__":
    main()