2.3.4. Random Forest Flowchart

Figure 6 shows the flow diagram of the whole random forest model. It is a collection of some decision trees. The process is the same as the decision tree. It preprocesses the data and selects some random samples from the dataset for training. For every selected sample, it forms a decision tree. First, the random forest model has been trained without fine-tuning. Then, just like the SVM, grid search has been used with 5-fold cross-validation and different parameter combinations such as the number of trees in the random forest (n_estimators), what function to use for the number of features to consider at every split, levels in the tree, and method of selecting samples for training each tree. To measure the quality of the tree, the Gini criterion has been used. The entropy criterion has also been tried in the model, but Gini criterion provides better accuracy.

Flowchart of random forest.