I computed a random forest model after determining the optimized parameters of ntree and mtry, than I evaluated its performance in the train and on the test. I obtained the following results: On the train: Sensibility: 78% Sensitivity: 81% Accuracy : 80%

On the test set: Sensibility: 98% Sensitivity: 99% Accuracy : 99%

How is this possible? Is that means that my model is good? How can I interprete these results.

Thank you in advance for your help



This is difficult to answer without the data but imagine, there is an underlying concept in your data that the random forest has discovered and is good at making predictions on.

Now imagine, there are some outliers or special cases or strangely behaving values and they all went into your training set just by chance. Then the forest should be better at the better behaved test set. Of course, the probability of this happening is larger with smaller training and test sets.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.