A 14-NN model is a type of "k nearest neighbor" (k-NN) algorithm that is used to estimate or predict the outcome of a mathematical query point based on 14 nearest neighbors. The k-NN algorithm is a nonparametric model typically used in regression and classification techniques. It is considered one of the simpler machine learning algorithms.
A 14-NN model is used to classify a new object based on 14 known objects. For example, if modelers want to know whether a query point is positive or negative, they look to the nearest neighbors for guidance. As the number of nearest neighbors enters the model, the outcome changes until there are enough examples to offer a good estimate. A larger number of nearest neighbors will help reduce noise in the classification, but can make the boundaries less clear.
K-NN models can also be used in regression problems. In regression, a dependent outcome variable is predicted based on a set number of independent variables. When the k-NN model is used, the outcome of a specific query point is estimated by calculating the average of the outcomes of its nearest neighbors. For example, the 14-NN model will take the average of the 14 nearest neighbors.