Updates to my Python Code For Strava

Introduction

In an earlier post, I discuss my Python code found in my GitHub repository that downloads a user's Strava data and allows the user to ask questions of the data that can't be asked on Strava's website or app (as useful as those are). If you'd like to use that code, please read that earlier post to get started, especially the section on OAuth 2.0 authorization which is a necessary, one-time step to be able to access Strava's API.

In this entry, I briefly review how to use the code and discuss two new features.

Review

Once you've gone through the OAuth 2.0 process, to use the code create a StravaAnalyzer instance like this:

sa = strava.StravaAnalyzer()

The constructor will download all of your Strava data and store it locally in a few different CSV files (see my earlier post).

New Features

The predict_avg_speed() method now supports activity types other than bike rides. To predict the average speed of a 10 mile hike with 364 feet of elevation gain, for example, run the following:

sa.predict_avg_speed(elev_gain=364, distance=10, activity_type="Hike")

Make sure that the activity_type matches Strava's name for that activity.

As before, this method uses three different models to make three predictions about the average speed of a proposed route. Two use regression. The third uses the average speed of similar activities. Use the dist_fudge and elev_fudge factors to indicate what it means to be similar. The defaults are each 0.1 which means that an activity is similar if the distance and elevation are within plus or minus 10% of those given.

suggest_similar_activities() is a new method that, given an elevation gain and a distance, returns a list of URLs of your past activities that have similar elevation and distance profiles. As with predict_avg_speed(), use the fudge factors to indicate what "similar" means. For example, if you want to see past bike rides that have an elevation gain of around 2000 feet and a distance of around 50 miles run the following:

sa.suggest_similar_activities(elev_gain=2000, distance=50, elev_fudge=0.1, dist_fudge=0.1, activity_type="Ride")

If you want greater variation, increase the fudge factors. You can also leave either distance or elevation gain out (but not both). For example, this call sa.suggest_similar_activities(distance=5, activity_type="Walk") will return all of your past walks on Strava that were around 5 miles long.

I find suggest_similar_activities() useful in planning my bike rides. I might know that I want to aim for a particular length ride with a certain amount of elevation gain. Given that information, I can look at what I've done in the past and decide if I want to ride any of those routes again. I can also use them as starting out points for new routes.