Penalized regression on principal manifolds with application to combustion modelling

Einbeck, J.; Isaac, B.; Evers, L.; Parente, A.

Abstract

For multivariate regression problems featuring strong and non–linear dependency patterns between the involved predictors, it is attractive to reduce the dimension of the estimation problem by approximating the predictor space through a principal surface (or manifold). In this work, a new approach for non- parametric regression onto the fitted manifold is provided. The proposed penal- ized regression technique is applied onto data from a simulated combustion sys- tem, and is shown, in this application, to compare well with competing regression routines.

Citation

Einbeck, J., Isaac, B., Evers, L., & Parente, A. (2012, December). Penalized regression on principal manifolds with application to combustion modelling. Presented at International workshop on statistical modelling, Prague

• Presentation Conference Type: Conference Paper (published)
• Conference Name: International workshop on statistical modelling
• Publication Date: Jan 1, 2012
• Deposit Date: May 29, 2013
• Publicly Available Date: Jun 19, 2013
• Volume: 1
• Pages: 117-122
• Book Title: 27th International Workshop on Statistical Modelling, 16-20 July 2012, Prague, Czech Republic ; proceedings.
• Keywords: Smoothing, Principal component analysis, Local principal manifolds,
• Public URL: https://durham-repository.worktribe.com/output/1156424
• Publisher URL: http://www.statmod.org/workshops_archive_proceedings_2012.htm
• Additional Information: http://www.maths.dur.ac.uk/~dma0je/Papers/einbeck_etal_iwsm2012.pdf