distribution for regression problems??
I have a dataset where the continuous target variable (regression) is very skewed. This affects the results when evaluating/testing. I googled the question yet I found out that most current methods are for classification problems (such as SMOTE, under/oversampling). One technique called SMOGN yet was not effective. Any suggestions/ideas? Thanks 🙏🏼
Try some transformation of the target column, like log.
Imbalance is found only with a classification problem rather there are number of techniques to transform your features to be normally distributed some of them are 1. Log transformation 2. Square root transformation 3. Boxcox transformation You can google to find more information about them
Обсуждают сегодня