k-nearest neighbors

The k-nearest neighbor (k-NN) classification is one of the easiest classification methods to understand (particularly when there is little or no prior knowledge about the distribution of the data). The k-nearest neighbor classification has a way to store all the known cases and classify new cases based on a similarity measure (for example, the Euclidean distance function). The k-NN algorithm is popular in its statistical estimation and pattern recognition because of its simplicity.

For 1-nearest neighbor (1-NN), the label of one particular point is set to be the nearest training point. When you extend this for a higher value of k, the label of a test point is the one that is measured by the k nearest training points. The k-NN algorithm is considered to be a lazy learning algorithm because the optimization is done locally, and the computations are delayed until classification.

There are advantages and disadvantages of this method. The advantages are high accuracy, insensitive...