The RF classification analysis was performed using R-packages “randomForest” [101], “caret” [102], and “randomForestExplainer” [103]. A RF classifier consists of collection of tree-structured classifiers where each tree casts a unit vote for a class/group for each set of predictor variables [36]. A growing number of studies in computational biology are using RF because of several advantages of the method. According to Qi [104], the RF method is not only nonparametric, but is interpretable and efficient. Further, the RF method can be applied to data with small sample size, multi-dimensional variables, and multiple layers/levels without compromising its prediction accuracy [104]. In a large-scale benchmark experiment, the RF algorithm was found to perform better than logistic regression in terms of prediction accuracy [105]. The two main parameters of the RF algorithm are the number of trees in the ensemble and the number of variables randomly selected for the splitting decision at each node. Two levels of randomness are used by the RF to construct the ensemble of trees: first, the model trains itself using a training data for creating each tree based
