Active Learning on Defect Paper

I ran all the data sets through weka's Bayes and Random Forest learners. The g-values produced by weka are very close to the g-values given by my implementation, but I still kept the g-values of weka. The figures are placed together in an under-construction version of the paper.