I am starting to use the accord framework which looks fantastic (and will hopefully allow me to avoid learning python). I finally got the C4.5 decision tree to work with discrete variables (although the tree is not picturing the output class properly in the tree view). However I am not getting anywhere near the same results as the C4.5 tree I get using WEKA on the same file. With WEKA I get 15% accuracy and less that 3% with ACCORD on the same data.
I saw elsewhere a note from 2013 that the decision tree was in beta. Can someone please confirm that the C4.5 algorightm is working with discrete variables before I spend more time on this.
Also you may want to note that there is a file missing (nuget.targets) when downloading the sample.
Thank you.
Stamati Crook
www.redware.com