Hello,

If I am using the KNN algo for a classification problem like the digit recognizer problem in Kaggle where the train data has 0-9 labels,should my k value be equal to 10??

I guess the value can’t be more than or less than 10 and has to be exactly equal to 10.Please correct me if I am wrong.

Also should I divide the dataset into say 4 samples containing 5 labels each and then train separate KNN’s keeping the k = 5 and combine them using ensemble methods??

I am currently running KNN on the whole dataset of about 29000 rows and 10 labels and it has been an hour since it started.

Can somebody kindly guide me here.