How to flatten percentage distribution across data?

Question

numpy

Hello colleagues. I have a question. There is a dataset. It has signs (columns) binary. but in these columns the ratio of True to False is not very even. For example True 90% and False 10%. How to align all the features of this data at once? By adding randomly new ones? Studio ML has such a tool called SMOTE, but it only works with one feature. Maybe there are some mechanisms for pandas or numpy? Thanks

Answer 1

The idea of ​​deconstructing real data with fake data to get something meaningful is a crazy idea.
Tools for the implementation of crazy ideas can be picked up, yes.

Answer 2

contrib.scikit-learn.org/imbalanced-learn/stable
But in general 90/10 is a healthy ratio for most cases.