I want to create a hybrid model that predicts heart disease. But before I start, I have a few questions:
Can the output of one classifier be used as input to another classifier? For example, can we use SVM->ANN->RF->LR? I think it's possible, but the label/target data may get reduced when passed from one classifier to another. Could you suggest some papers or codes related to this?
What parameters should I consider when building a hybrid model, apart from Kappa value?
How should I select algorithms for a hybrid model?
What are some best practices or common methods for building a hybrid model, apart from ensemble learning? Any suggestions would be appreciated.