DataTaunew | comments | leaders | submitlogin
4 points by jcbozonier 3672 days ago | link | parent

Whoa wait. His rationale for 5 teams being outliers seems pretty weak. If I'm reading this right the outliers make up around 17% his data points as well. Seems like a lot of data to disregard as "outliers".

This seems more like data shaping than noise reduction.



1 point by roycoding 3672 days ago | link

Seems like a fine line. A more robust inspection of distribution of stars (and anti-stars?) on the teams would make the argument more convincing.

This definitely falls in the "fun, back-of-the envelope" category versus "serious" academic research. Both are worthwhile.

I definitely enjoyed reading this analysis. Always nice to see the code and reasoning.

-----

2 points by jcbozonier 3671 days ago | link

I agree! It was super easy to follow. I agree, rather than just throwing out the outliers, it'd be great to dive into why they might be different.

-----




RSS | Announcements