Machine Learning Lesson of the Day: The K-Nearest Neighbours Classifier
February 21, 2014 Leave a comment
- identifying the K data (the K “neighbours”) in the training set that are closest to
- counting the number of “neighbours” that belong to each class of the target variable
- classifying by the most common class to which its neighbours belong
K is usually an odd number to avoid resolving ties.
The proximity of the neighbours to is usually defined by Euclidean distance.