So I'm kinda new to machine learning and I am trying to learn as I build my project. I'm building a classification model for sleep disorders using Voting Ensemble and I have three base models: Logistic Regression, Random Forest and SVM. Now I want to combine these models using a Voting Ensemble with soft voting.I don't really have anyone to ask so I came here to ask about the correct approach: I know LR and SVM needs to be scaled but because they all need to be in the same format (as in scaled or raw), I need to scale the rf model too right? I know rf doesn't need scaling but others do, what is the best approach here? (I did my best to explain my problem, I hope it is understandable as English isn't my first language.) I tried both approaches: training Random Forest on scaled data to match other models, and also manually combining predictions where each model receives its appropriate data format. I expected the second approach to perform better since Random Forest shouldn't need scaling, but wanted to verify this is the correct practice when using VotingClassifier. Continue reading...
