I am trying to solve the hand written digit recognizer problem of Kaggle.I am trying to compare the error rates of various models as below:
As can be seen for the naiveBayes model there is an error.I understand that this is happening because nb returns posterior probabilities.I am sure there is a way to specify that I want predicted class instead of prob but how do I do it in the errorest function??
Can someone please help me with this??