0
Follow
0
View

Improving performance result of classification for severely imbalance data having abnormal skewed distribution

daiyuanrui 注册会员
2023-01-25 14:48

Have you tried using genuine oversampling?

Your questions states you are oversampling by using an undersampling method, which will remove data points leaving less data for training which may be why your model is struggling.

I believe the oversampling functionality is built into imblearn so should be a quick experiment to try it.

https://imbalanced-learn.org/stable/references/generated/imblearn.over_sampling.RandomOverSampler.html