[discussion] data preprocessing before or after splitting

It could be a very newbie question.

I have a dataset with financial values for companies , some features have missing values and the data is unbalanced.

Here's what i did

Scaling the Data with sklearn standard Scaler
Missing values imputation with knn imputer and iterative imputer trying several strategies and choosing the best one
Oversampling with smote

Testing with random forest classifier.


The next step is to add more features :
1 by calculating some features from my already …

data data preprocessing machinelearning