Multivariate regression smoothing through the 'fallling net'

Taylor, James; Einbeck, Jochen

Abstract

We consider multivariate regression smoothing through a conditional mean shift procedure. By computing local conditional means iteratively over a set or grid of target points, at each iteration a `net' is formed which gently drifts towards the data cloud, until it settles at the conditional modes of the response distribution. The method is edge-preserving and allows for multi-valued response.

Citation

Taylor, J., & Einbeck, J. (2011, December). Multivariate regression smoothing through the 'fallling net'. Presented at 26th international workshop on statistical modelling., Valencia

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 26th international workshop on statistical modelling.
• Publication Date: Jul 11, 2011
• Deposit Date: Aug 23, 2011
• Publicly Available Date: Jun 14, 2013
• Pages: 597-602
• Book Title: 26th International Workshop on Statistical Modelling, 5-11 July 2011, Valencia, Spain ; proceedings.
• Keywords: Conditional density, Modal regression, Smoothing.
• Public URL: https://durham-repository.worktribe.com/output/1157737
• Publisher URL: http://www.statmod.org/workshops_archive_proceedings_2011.htm