Closed quique0194 closed 3 years ago
It seems like the format of the output is:
[avg(scoring) avg(number_of_features)] [std(scoring) std(number_of_features)] [min(scoring) min(number_of_features)] [max(scoring) max(number_of_features)]
And -10000
is the score you get when the number of selected features is greater than the number of max_features
you set.
So what is the purpose of setting max_features?
Sorry for the extremely late reply.
-10000
is a "bad score" that get assigned to invalid feature combinations to discourage their selection (combinations with a number of features > max_features
are invalid).
It looks like, in your case, all the generated combinations are invalid (number of features > max_features
). Commit 3d62422 should improve the situation by generating only valid combinations for the initial population.
Please feel free to reopen the issue if necessary.
I have the following output, Could you explain in the documentation what is the meaning of each column? Thanks!