Determining k in k-means clustering by exploiting attribute distributions

Bocking, Oscar (2018) Determining k in k-means clustering by exploiting attribute distributions. Bachelor's Thesis, Artificial Intelligence.

Preview	Text AI_BA_2018_BockingO.pdf Download (566kB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (94kB)

Abstract

Methods for estimating the natural number of clusters (k) in a data set traditionally rely on the distance between points. In this project, an alternative was investigated: exploiting the distribution of informative nominal attributes over the clusters with a chi-squared test of independence, to see which value of k partitions the data in a way that is least likely to be random. Artificial data sets are used to assess the strategy's performance and viability in comparison to a well-established distance-based method. Results indicate that the proposed strategy has a tendency to overestimate k, and only performs consistently with some types of attribute. Despite this, it has value as a heuristic method when attributes are available due to non-reliance on distance information.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Schomaker, L.R.B.
Degree programme:	Artificial Intelligence
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	13 Jul 2018
Last Modified:	20 Jul 2018 11:36
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/17848

Actions (login required)

View Item