the percentage of data to "hold out" for pruning
the information gain metric to use ("entropy" or "multinomial")
for each node, the fraction of features to randomly consider as potential splitting features
the maximum desired depth of the trained decision tree
Factory constructor of DecisionTree.
Factory constructor of DecisionTree.
Randomly splits data into training:validation (according to the validationPercentage param). Does reduced-error pruning on validation data. Uses a uniform prior over training labels where each label is assumed to have been seen once already. This Laplace smoothing affects the probability distribution over labels for each feature.
training and validation data
the induced decision tree
A DecisionTreeTrainer trains decision trees from data.