K means

#Terms

WHAT IS… ?

k-means clustering is a method of vector quantization, originally from signal processing, that aims to partition n observations into k clusters. The algorithm works by initially assigning each observation to its own cluster and then moving observations around among the clusters to improve the overall fit, until no observations change clusters. The resulting assignment of observations to clusters is called a cluster assignment or a cluster analysis.

HOW IS IT USED ?

The k-means algorithm is a method to cluster data points in an n-dimensional space. The algorithm partitions n observations into k clusters in which each observation belongs to the cluster with the nearest centroid, serving as a prototype of the cluster. A good choice for k typically depends on the noise level of the data and the intrinsic dimensions of the problem.