Testability of challenges recommended by the recommender api

[SathyaJayabal]

How can we test the correctness of the challenges that are recommended by the recommender api ?

[g0rynich]

1. Simple option. Pick 3-5 distinct skill profiles (design regular, marathon regular, soft-dev regular...) and check if sorting brings up related tracks.
2. Complex option. We can generate a dataset of a) Skill profiles at the given moment of the past; b) open challenges at the given moment of the time; Who actually ended up participating in each of the challenges. And check if the challenges the members ended up participating in - are ranked high.