Decision Tree ensemble algorithms

For now library offers two algorithms - bagging and random forest.

Bagging

Usage bagging is similiar to usage of decision tree learning algorithms:

 const options = {
            treesCount: 200,
            subsetItemsCount: 100,
            learningMethod: 'C45',
        };
        const baggingAlgorithm = new Bagging(options);
        trees = baggingAlgorithm.buildTreesBag(trainingSet, labels);

treesCount - number of tree in set

subsetItemsCount - number of items in learning set per one decision set

learningMethod - method for learning decision tree

Full example of usage you can find here.

Random Forest

Usage bagging is similiar to usage of bagging:

 const options = {
            treesCount: 200,
            subsetItemsCount: 100,
            featureSubset: 5,
            learningMethod: 'C45',
        };
        const randomForestAlgorithm = new RadnomForest(options);
        trees = baggingAlgorithm.buildTreesBag(trainingSet, labels);

treesCount - number of tree in set

subsetItemsCount - number of items in learning set per one decision set

featureSubset - number of random labels which will be use for splitting in single tree

learningMethod - method for learning decision tree

Full example of usage you can find here.