[MATH-1524] "chooseInitialCenters" should move out from KMeansPlusPlusClusterer

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Assignee: Unassigned
• Reporter: Chen Tao
• Created: 10/Mar/20 07:58
• Updated: 30/May/21 13:31

Description

There are two reason for "chooseInitialCenters" should be move out from "KMeansPlusPlusClusterer":

1. k-means++ clusterer is a special case of k-means clusterer, that k-means++ initialize the cluster centers with k-means++ algorithm. Another case is initialize the cluster centers with random points.
2. The mini batch k-means will reuse "chooseInitialCenters".

Issue Links

• Blocked: MATH-1509 Implement the MiniBatchKMeansClusterer
• links to: GitHub Pull Request #125
• links to: GitHub Pull Request #130
• links to: GitHub Pull Request #131
• links to: GitHub Pull Request #133