In the regression prediction model, the same as the model. In terms of classification (Ensemble method), can we use a combination of several different models? If yes, how?
Your post is confusing, and if you want proper answers it may help to edit it. My understanding is that you are asking whether the ensembling works for regression like it does for classification. If so, the answer is yes, and the approach is the same. We make several regression models, and feed their predictions and the ground truth into a final model which will learn how much each individual model should contribute. If you will, the final model learns weight coefficients for each individual model.