Hands-On Recommendation Systems with Python

k-nearest neighbors

k-nearest neighbors (k-NN) is perhaps the simplest machine learning algorithm. In the case of classification, it assigns a class to a particular data point by a majority vote of its k nearest neighbors. In other words, the data point is assigned the class that is the most common among its k-nearest neighbors. In the case of regression, it computes the average value for the target variable based on its k-nearest neighbors.

Unlike most machine learning algorithms, k-NN is non-parametric and lazy in nature. The former means that k-NN does not make any underlying assumptions about the distribution of the data. In other words, the model structure is determined by the data. The latter means that k-NN undergoes virtually no training. ...