aartgoossens / athletic_pandas

Extension of Pandas that provides workout analysis tools

Athletic Pandas

---

DEPRECATION NOTICE: This library has been deprecated. Code has been moved to sweatpy and development will continue there.

Introduction

Athletic Pandas is an extension of pandas designed to make workout analysis a breeze. The current state of the project is very beta: features might be added, removed or changed in backwards incompatible ways. When the time is right a stable version will be released. Get in touch with me or create an issue if you have problems/questions/feature requests/special use cases.

Installation

This library can be installed via PyPI:

pip install athletic_pandas

Usage

The core feature of this library is the WorkoutDataFrame. This class is a subclass of pandas.DataFrame and offers all the functionality that is available for this class, plus some workout specific functionality.

After installation, the WorkoutDataFrame can be imported this way:

from athletic_pandas.models import WorkoutDataFrame

To load some example data from this repository:

import pandas as pd
wdf = WorkoutDataFrame(pd.read_csv('./tests.example_files/workout_1.csv'))

Basic operations on the data are provided by pandas.Dataframe superclass:

wdf.power.mean()
>> 131.65231973169369
wdf.power.max()
>> 280
wdf.heartrate.plot()

For some workout specific functionality information about the athlete is needed. This can be set on the WorkoutDataFrame as follows:

from athletic_pandas.models import Athlete
wdf.athlete = Athlete(name='Chris Froome', cp=175, w_prime=20000)

When this is done for example the W'balance can be computed, added to the wdf and plotted:

wdf = wdf.assign(w_prime_balance=wdf.compute_w_prime_balance())
wdf.w_prime_balance.plot()

It is also possible to compute and plot the mean max power for this workout:

mean_max_power = wdf.compute_mean_max_power()
# Get the mean max power for 60 seconds
mean_max_power[60]
>> 215.03278688524591
mean_max_power.plot()

Another feature of this library is that you can create custom workout and for example analyze the W'balance of it:

wdf = WorkoutDataFrame(dict(
    power=[100]*60 + [200]*120 + [100]*60
))
wdf.athlete = Athlete(cp=175, w_prime=20000)
wdf = wdf.assign(w_prime_balance=wdf.compute_w_prime_balance())
wdf.w_prime_balance_min()
>> 17320.207532891662

There are already more algorithms available on the WorkoutDataFrame such as alternatives for the Skiba W'balance algorithm. More algorithms will be added later.

This documentation is quite compact at the moment but will be extended with more examples in the near future. Stay tuned!

Contributing

Coming soon...

Contributors

Aart Goossens

License

See LICENSE file.

athletic_pandas's Issues

Add Athlete model

Add Athlete that serves as a class attribute for Workout to provide additional data like FTP.

Add CalenderDataFrame

Add model CalenderDataFrame that enables longitudinal analysis of e.g. training load.
Preferably stores WorkoutDataFrame objects in a column called 'data'

test

test

Add data validation on Workout initialization

Upon initialization of a Workout the data needs to be verified. E.g.:

• Is data sampled at 1Hz
• Are values in the expected format

Add workout_type attribute to workout

For workout validations. E.g. grade_adjust_pace can only be calculated for running workouts.
Types: cycling, running, swimming

Add isort to Makefile

Add isort command to Makefile to make checking import easier