On the advantages of the non-concave penalized likelihood model selection method with minimum prediction errors in large-scale medical studies

Karagrigoriou, A; Koukouvinos, C; Mylona, K

dc.contributor.author	Karagrigoriou, A	en
dc.contributor.author	Koukouvinos, C	en
dc.contributor.author	Mylona, K	en
dc.date.accessioned	2014-03-01T01:34:01Z
dc.date.available	2014-03-01T01:34:01Z
dc.date.issued	2010	en
dc.identifier.issn	0266-4763	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/20644
dc.subject	Deviance	en
dc.subject	Generalized linear model	en
dc.subject	Highdimensional data set	en
dc.subject	Model selection	en
dc.subject	Non-concave penalized likelihood	en
dc.subject	Trauma	en
dc.subject.classification	Statistics & Probability	en
dc.subject.other	ASYMPTOTICALLY EFFICIENT SELECTION	en
dc.subject.other	VARIABLE SELECTION	en
dc.subject.other	CROSS-VALIDATION	en
dc.subject.other	REGRESSION	en
dc.subject.other	ORDER	en
dc.subject.other	LASSO	en
dc.title	On the advantages of the non-concave penalized likelihood model selection method with minimum prediction errors in large-scale medical studies	en
heal.type	journalArticle	en
heal.identifier.primary	10.1080/02664760802638116	en
heal.identifier.secondary	http://dx.doi.org/10.1080/02664760802638116	en
heal.language	English	en
heal.publicationDate	2010	en
heal.abstract	Variable and model selection problems are fundamental to high-dimensional statistical modeling in diverse fields of sciences. Especially in health studies, many potential factors are usually introduced to determine an outcome variable. This paper deals with the problem of high-dimensional statistical modeling through the analysis of the trauma annual data in Greece for 2005. The data set is divided into the experiment and control sets and consists of 6334 observations and 112 factors that include demographic, transport and intrahospital data used to detect possible risk factors of death. In our study, different model selection techniques are applied to the experiment set and the notion of deviance is used on the control set to assess the fit of the overall selected model. The statistical methods employed in this work were the nonconcave penalized likelihood methods, smoothly clipped absolute deviation, least absolute shrinkage and selection operator, and Hard, the generalized linear logistic regression, and the best subset variable selection. The way of identifying the significant variables in large medical data sets along with the performance and the pros and cons of the various statistical techniques used are discussed. The performed analysis reveals the distinct advantages of the non-concave penalized likelihood methods over the traditional model selection techniques. © 2010 Taylor & Francis.	en
heal.publisher	ROUTLEDGE JOURNALS, TAYLOR & FRANCIS LTD	en
heal.journalName	Journal of Applied Statistics	en
dc.identifier.doi	10.1080/02664760802638116	en
dc.identifier.isi	ISI:000272848000002	en
dc.identifier.volume	37	en
dc.identifier.issue	1	en
dc.identifier.spage	13	en
dc.identifier.epage	24	en