Explains how splitting data into a classifier and training and test sets can help determine K (the number of similiar data points to be considered).