Previously I used Opencv random forest classifier for my data. But problem with Opencv is posterior probabilities. It is just for binary classification and does not support more than two classes probabilities (CvRTrees::predict_prob opencv function). While the number of classes I have in my data is 7. But ALGLIB support more than two probabilities (alglib::dfprocess). I used this function to get probabilities for my seven classes. But problem with me is now the results I obtained with Alglib are much poor than OPencv classifier.
Can anybody tell me how can I solve my problem ? I need to get probability prediction for all the seven classes which I am using.
