Truecluster

Truecluster is a framework for robust, scalable clustering with model selection. Its central purpose is to help users determine the optimal number of clusters objectively, including when working with non-parametric clustering algorithms. The project addresses a long-standing challenge in clustering: deciding how many groups a dataset contains. The website presents Truecluster as a method intended to make that choice more systematic, rather than relying solely on an algorithm’s clustering output.

The site provides publications describing the framework and a related approach called Truecluster matching, both credited to J. Oehlschlägel and listed as available through arXiv. Visitors can also explore examples and example data to see the methods in context. The material is presented as a research and technical resource, with contact available by email for inquiries. Truecluster’s stated focus is the development and use of robust clustering methods that combine cluster formation with model selection. Its website also includes terms of use, privacy information, and an imprint.