News

For decision tree classification, the variable to predict is most often ordinal-encoded (0, 1, 2 and so on) The numeric predictors do not need to be normalized to all the same range -- typically 0.0 ...