title	kmeans
date	2018-12-11 15:00:00 +0200
layout	default
published	true

K-Means

Those operators perform the unsupervised learning algorithm K-Means. It is used to cluster the data in K homogeneous and provide a 2-dimensional synthetic visualization of the result. For each of the K groups, an "artificial mean point" called the centroid is built and displayed in the visualisation.

There are three different K-means algorithms in IKATS, depending on the nature of the inputs:

K-Means on SAX

Perform K-Means algorithm on SAX words. It is dedicated to be applied to the output of the SAX operator

K-Means on Patterns

Perform K-Means algorithm on patterns. It is dedicated to be applied to the output of the Random Projections operator.

K-Means on Time Series

Perform K-Means algorithm directly on time series data.

Return to the list of all operators