Assume that a large number of binary features are added to a dataset with two class labels c1 and c2, such that for each added feature f, the class conditional probability P(f = 0|c1) = P(f = 0|c2). What effect will the addition of such features have on the accuracy of naive Bayes and kNN respectively?

I calculated for Naive Bayes P(c1| f1=0,..,fn=0) and P(c2| f1=0,..,fn=0) obtaining then that are the same as relative frequences of both classes, but how could this affect the accuracy?


Your Answer

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

Browse other questions tagged or ask your own question.