UCL logo

UCL Discovery

UCL home » Library Services » Electronic resources » UCL Discovery

Fixed point clusters for linear regression: Computation and comparison

Hennig, C; (2002) Fixed point clusters for linear regression: Computation and comparison. Journal of Classification , 19 (2) pp. 249-276. 10.1007/s00357-001-0045-7.

Full text not available from this repository.

Abstract

A subset of a regression dataset, i.e., consisting of an independent variable and one or more regressors, is called regression Fixed Point Cluster (FPC) if it reproduces itself under the following procedure: Its linear regression and variance estimators are computed, all points too far from the regression hyperplane are declared as outliers, and the subset under consideration is exactly the set of non-outliers w.r.t. itself. In this paper an algorithm is developed, which aims to find all FPCs of a dataset corresponding to well separated linear regression subpopulations. Its ability to find such subpopulations under the occurrence of outliers is compared to methods based on ML-estimation of mixture models by means of a simulation study. Furthermore, FPC analysis is applied to a real dataset.

Type: Article
Title: Fixed point clusters for linear regression: Computation and comparison
DOI: 10.1007/s00357-001-0045-7
UCL classification: UCL > School of BEAMS
UCL > School of BEAMS > Faculty of Maths and Physical Sciences
URI: http://discovery.ucl.ac.uk/id/eprint/148270
Downloads since deposit
0Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item