我需要删除特征之间的多重共线性

Question

I have categorical variables such as Gender,Anxiety,Alcoholic and when i convert these categorical variables into numerical values using encoder techniques then all these variables resembles same in values and then multi co linearity is existing. How i can convert these variables to number so that multi co linearity doesn't exist. All three variables are important for prediction of target variable.

Answer 1

You don't need to transform the data.Instead you can change the way that you are calculating correlation between variables. As these are categorical features, you have to use Chi-Squared test of independence.Then, you won't be facing this issue.