3 ms·
There is this famous paper "I Just Ran Two Million Regressions". Any half decent statistician will tell you what's wrong with that type of approach.
by lottin 9y ago
There is this famous paper "I Just Ran Two Million Regressions". Any half decent statistician will tell you what's wrong with that type of approach.
- marcosdumay 9y agoIn machine learning circles, this is called "overfitting".
- joe_the_user 9y agoCould one describe machine learning as "combine enough bad statistical tests that the cries of 'it seems to work' from managers drown out all the 'it's wrong' cries from the statisticians"