Cost sensitive discretization of numeric attributes

摘要: Many algorithms in decision tree learning are not designed to handle numeric valued attributes very well. Therefore, discretization of the continuous feature space has be carried out. In this article we introduce concept cost sensitive as a preprocessing step induction classifier and an elaboration error-based method obtain optimal multi-interval splitting for each attribute. A transparant description steps involved is given. We also evaluate its performance against two other well known methods, i.e. entropy-based pure on real life financial dataset. From algoritmic point view, show that important deficiency from methods can solved by introducing costs. application discovered using recommended. To conclude, use ROC-curves illustrate under particular conditions cost-based may optimal.