Using mixed models to smooth regression coefficients corresponding to inter-correlated explanatory variables: an application to weather covariates

Michelle Sims, D A Elston, A Larkham

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

A linear mixed model version of penalised regression is proposed. We
demonstrate how to estimate the effects and standard errors of a set of intercorrelated explanatory variables by imposing penalties on the regression coefficients by modelling them as random effects. An example is given using correlated weather covariates.
Original languageEnglish
Title of host publicationStatistical Solutions to Modern Problems. Proceedings of the 20th International Workshop on Statistical Modelling
EditorsA R Francis, K M Matawie, A Oshlack, G K Smyth
Place of PublicationSydney, Australia
PublisherStatistical Modelling Society
Pages447-451
ISBN (Print)1 74108 101 7
Publication statusPublished - 2005
Event20th International Workshop on Statistical Modelling - Sydney, Australia
Duration: 10 Jul 200515 Jul 2005

Conference

Conference20th International Workshop on Statistical Modelling
CountryAustralia
CitySydney
Period10/07/0515/07/05

Fingerprint Dive into the research topics of 'Using mixed models to smooth regression coefficients corresponding to inter-correlated explanatory variables: an application to weather covariates'. Together they form a unique fingerprint.

  • Cite this

    Sims, M., Elston, D. A., & Larkham, A. (2005). Using mixed models to smooth regression coefficients corresponding to inter-correlated explanatory variables: an application to weather covariates. In A. R. Francis, K. M. Matawie, A. Oshlack, & G. K. Smyth (Eds.), Statistical Solutions to Modern Problems. Proceedings of the 20th International Workshop on Statistical Modelling (pp. 447-451). Statistical Modelling Society.