Bag of Little Random Forests (BLRF)
Random Forests are an ensemble method that utilizes a number of decision trees to make robust predictions in both regression and classification settings. However, the process of bootstrap aggregation, the mechanism underlying the Random Forests algorithm, requires each decision tree to physically store and perform computations on data sets of the same size as the training data set. This situation is oftentimes impractical given the large size of data sets nowadays. To address this problem, we introduce the Bag of Little Random Forests (BLRF), a new algorithm that combines the Random Forests with the Bag of Little Bootstraps resampling method, aiming to achieve a better computational profile while producing predictions with comparable accuracy as those of the Random Forests.