ADVANCES IN ROBUST CLUSTERING FOR REGRESSION STRUCTURES
We investigate the properties of state-of-the-art robust clusterwise regression tools based on trimming and restrictions, under a variety of data configurations and contamination schemes. The data are generated through a new flexible simulation tool for mixtures of regressions, where the user can control the degree of overlap between the groups. The tool is also used to illustrate the effect of concentrated “noise type” contamination on the fit of robust clusterwise regression methods.
PERROTTA Domenico;
TORTI Francesca;
CERIOLI Andrea;
RIANI Marco;
2017-12-19
Universitas Studiorum S.r.l. Casa Editrice
JRC106435
978-88-99459-71-0,
http://www.universitas-studiorum.it/1/cladag_2017_book_of_short_papers_2852494.html,
https://publications.jrc.ec.europa.eu/repository/handle/JRC106435,
Additional supporting files
| File name | Description | File type | |
|---|
