2.4.2 Random forest (RF)
Breiman improved the regression tree model based on the Bagging algorithm and proposed the random forest algorithm (BREIMAN, 2001) , which consists of sub-training sets and sub-regression models (decision trees), which extracts m multiple sample data points from the original sample set D through Bootstrap resampling method to form a sub-training sample set with the same sample size as the original one (see Figure 5 (b)). For each sub-training sample set, a sub-regression model is constructed, which is called random forest model (Das et al., 2017; Ibarra-Berastegi et al., 2015; Nashwan and Shahid, 2019) .