You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi,
I see in the project front page that the continuous data should be discretized. But:
1- When I give it continuous data I do not get any error.
2- If I want to discretize data, is it sufficient to just run for example sklearn's KBinsDiscretizer and give it the required number of bins with uniform strategy? This will not be the same as the strategy mentioned in the paper (i.e. considering mean+/-std and binning to -1, 0, and 1 accordingly) but I do not think that really matters. If the feature values are changed by a scale, why the feature selection mechanism should give us different results?
I hope to hear from you soon.
Best Regards
Shahriar
The text was updated successfully, but these errors were encountered:
Hi,
I see in the project front page that the continuous data should be discretized. But:
1- When I give it continuous data I do not get any error.
2- If I want to discretize data, is it sufficient to just run for example sklearn's KBinsDiscretizer and give it the required number of bins with uniform strategy? This will not be the same as the strategy mentioned in the paper (i.e. considering mean+/-std and binning to -1, 0, and 1 accordingly) but I do not think that really matters. If the feature values are changed by a scale, why the feature selection mechanism should give us different results?
I hope to hear from you soon.
Best Regards
Shahriar
The text was updated successfully, but these errors were encountered: