What is a Ensemble Model? (Stacking)

preview_player
Показать описание


Рекомендации по теме
Комментарии
Автор

I read on cross validation for time series that actually some samples should optimally be „learn-only“ to avoid information leakage. This poses a problem to the ensemble-learner, as it does not learn on the full sample set, but only on those samples predicted in the CV. Any thoughts on that?

BTW for me as someone who worked with time series a lot already I am debating seeing your course but I fear it’s too expensive for the surplus information I may be taking from it...

BlueQualityRhythm
Автор

Hello, your presentation is awesome. I have just one question. When we got 4 columns of prediction, then we need to divide that data again into training and testing for meta learners algorithms? or we just train but the entire predictions and predict by the same data. Highly appreciated

engr.abdullahazzamsafi
Автор

Is there an advantage in RMSE- over handling the same methods (ARIMA, GLMNET, XGBOOST SVM) separately?

hansmeiser
visit shbcf.ru