K-Nearest Neighbors is a supervised learning algorithm that makes a prediction based on the K nearest neighbors of a target sample. In case of classification, the class with the most votes among those neighbors is assigned to the sample. In case of regression, the average label of the neighbors is assigned.