Ensembling by averaging
Within the context of regression models, the predictions are the numeric values of the variables of interest. Combining the predictions of the output due to the various ensemblers is rather straightforward; because of the ensembling mechanism, we simply interpret the average of the predicted values across the ensemblers as the predicted value. Within the context of the classification problem, we can carry out simple averaging and weighted averaging. In the previous section, the ensemble had homogeneous base learners. However, in this section, we will deal with heterogeneous base learners.
We will now consider a regression problem that is dealt with in detail in Chapter 8, Ensemble Diagnostics. The problem is the prediction of housing prices based on over 60 explanatory variables. We have the dataset in training and testing partitions, and load them to kick off the proceedings:
> # Averaging for Regression Problems
> load("../Data/ht_imp_author.Rdata"...
