It is really a big doubt.
First I made a base model and then I did a dimensionality reduction process. After that I made a model with reduction features.
Then, I was planning to do a hyperparameter tuning.
Question is what values i could use in the tuning process like 100% of the data or feature selection data?
The second Question is after completing the tuning process….we have to create the model right, which date is valid to use in the model? (100% or feature selection data)
Final Question, for the classification process we normally do a SMOTE right? the smote data is valid to use for all models or nor (if not, for which is valid to use)