Please sir answer...

After running OLS regression, bagging, and boosting, you know that you can use model comparison/screening to select the single best model or model averaging to combine the parameters of all models. Within the approach of model comparison there are many different ways and criteria. You can do a manual comparison based upon AICc and BIC, or a formal comparison based on Entropy R-square, generalized R-square, mean absolute deviation, RMSE, mis-classification rate, ROC curve, lift curve...etc. Between model comparison/screening and model averaging, which one will you use? Although there is no right or wrong answer, you need to explain your rationale, such as evaluating the pros and cons of each approach.