OpenText Analytics Database 26.2.x – Clustering algorithms

Data-Analysis: K-means

Mon, 01 Jan 0001 00:00:00 +0000

You can use the k-means clustering algorithm to cluster data points into k different groups based on similarities between the data points.

k-means partitions n observations into k clusters. Through this partitioning, k-means assigns each observation to the cluster with the nearest mean, or cluster center.

For a complete example of how to use k-means on a table, see Clustering data using k-means .

Data-Analysis: K-prototypes

Mon, 01 Jan 0001 00:00:00 +0000

You can use the k-prototypes clustering algorithm to cluster mixed data into different groups based on similarities between the data points. The k-prototypes algorithm extends the functionality of k-means clustering, which is limited to numerical data, by combining it with k-modes clustering, a clustering algorithm for categorical data.

See the syntax for k-prototypes here.

Data-Analysis: Bisecting k-means

Mon, 01 Jan 0001 00:00:00 +0000

The bisecting k-means clustering algorithm combines k-means clustering with divisive hierarchy clustering. With bisecting k-means, you get not only the clusters but also the hierarchical structure of the clusters of data points.

This hierarchy is more informative than the unstructured set of flat clusters returned by K-means. The hierarchy shows how the clustering results would look at every step of the process of bisecting clusters to find new clusters. The hierarchy of clusters makes it easier to decide the number of clusters in the data set.

Given a hierarchy of k clusters produced by bisecting k-means, you can easily calculate any prediction of the form: Assume the data contain only k' clusters, where k' is a number that is smaller than or equal to the k used to train the model.

For a complete example of how to use bisecting k-means to analyze a table, see Clustering data hierarchically using bisecting k-means.