Algorithm.
Random forest.
|
Initialization: A training set S: = {(xi,yi)}, T features, and number of trees in forest P 1. Select M trees from the dataset, in order to to construct a decision tree 2. Redo the previous step P times 3. At each node: 4. Construct a small subset of F, call it f 5. Separate the most appropriate features in f 6. The category that gains the majority votes will be given a new record The Output will be the selected features that have the highest accuracy score |