A report on experiments with weighted relative accuracy in CN2Ljupco Todorovski, Peter Flach, Nada Lavrac, A report on experiments with weighted relative accuracy in CN2. CSTR-00-003, Department of Computer Science, University of Bristol. March 2000. PDF, 202 Kbytes.
In this report, the prediction performances of two rule evaluation measures, accuracy and weighted relative accuracy, are compared. These two measures were used in two different versions of CN2 algorithm for inducing classification rules. The first version (CN2-acc) is the original one, where the accuracy measure for evaluating rules is used. The second one (CN2-wracc) uses weighted relative accuracy. The performance of two versions of CN2 was compared in sense of classification accuracy as well as number and length of the induced rules.