Analysis and extension of decision trees based on imprecise probabilities: Application on noisy data

摘要: An analysis of a procedure to build decision trees based on imprecise probabilities and uncertainty measures, called CDT, is presented. We compare this with the classic ones Shannon's entropy for precise probabilities. found that handling imprecision key part obtaining improvements in method's performance, as it has been showed class noise problems classification. present new building extending CDT's processing all input variables. show, via an experimental study data set general (noise variables), builds smaller gives better results than original CDT trees.