It is an interesting question. I used R to parse XML files (tcx file exported from endomondo) and then recovered the hundred and hundred GPS positions (latitude and longitude) for each training. A GPS position in the tcx file looks like this:

Once I had this information, I decided to develop an algorithm to compare trainings, basically comparing GPS positions. This algorithm receives a threshold. It is a value (%) to consider two trainings as similar.

The algorithm looks like this:

The output is similar to this:

[1] “The similitude between trainings 2016-01-04 and 2016-01-07 is 99.83035”
[1] “The similitude between trainings 2016-01-04 and 2016-01-11 is 99.7419255”
[1] “The similitude between trainings 2016-01-04 and 2016-01-18 is 99.8459195”
[1] “The similitude between trainings 2016-01-04 and 2016-02-24 is 99.7637684999999”
[1] “The similitude between trainings 2016-01-04 and 2016-03-17 is 99.7175975”
[1] “The similitude between trainings 2016-01-04 and 2016-03-31 is 99.5545275”

Then I generated an excel as output with the times I repeated a training. It is the input to Tableau. The map looks loke this:

In different red shades I can see my frequent trainings. Basically, what I supposed: from my job to home.