Bayesian Model Averaging

Hello!

I'm currently developing a forecasting model. My doubt is in the post processing part. I'm trying to find a model that, receiving several forecasts as input, gives as output the most probable or a combination of some of the most probable. Will the Bayesian Model Averaging work here? Or do you suggest another one that is more qualified?

Thank you for the attention!

RitaSantos said:

Hello!

I'm currently developing a forecasting model. My doubt is in the post processing part. I'm trying to find a model that, receiving several forecasts as input, gives as output the most probable or a combination of some of the most probable. Will the Bayesian Model Averaging work here? Or do you suggest another one that is more qualified?

Thank you for the attention!

Click to expand...

It may be that my understanding of Bayesian Model Averaging (BMA) is wrong but a (maybe too simplistic) explanation is that the average prediction of a set of predictions tend to do a better job of predicting than any of the single predictors as long as all of the predictors do at least a pretty good job. In that sense, BMA should work for your problem as long as the predictions (several forecasts) do a pretty good job by themselves.