Feature selection using decision trees in the problem of JSM classification
I. B. Yashkov · Automatic Documentation and Mathematical Linguistics · 2014
A method of feature selection is considered for training a classification algorithm based on the JSM method. The method is based on the use of decision trees. It includes construction of a maximal tree using the C4.5 algorithm and the preparation of a series of truncated trees based on the criterion of minimal cost complexity. The results and parameters of the JSM classifier with different numbers of selected features are compared.