I'm training a random forest classifier with the following command:
rusTree = fitensemble(trainX,trainY,'RUSBoost',1000,t,'LearnRate',0.1,'nprint',100);
I'm then curious about how well the model works, and ran the following commands:
L = loss(rusTree,testX,testY,'mode','cumulative');[label,score] = predict(rusTree,testX);
With label and the testY, I evaluated the amount of True positives and False positives in my data. Since L is the classification error, I expected that my evaluation and L would have some connection. However, they seem to be irrelevant. What exactly am I getting out of predict() and loss()?
Thank you!
June.
Best Answer