Comparing stochastic optimization methods for variable selection in binary outcome prediction, with application to health policy

Fouskakis, D; Draper, D

dc.contributor.author	Fouskakis, D	en
dc.contributor.author	Draper, D	en
dc.date.accessioned	2014-03-01T01:28:03Z
dc.date.available	2014-03-01T01:28:03Z
dc.date.issued	2008	en
dc.identifier.issn	0162-1459	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/18686
dc.subject	Bayesian decision theory	en
dc.subject	Cross-validation	en
dc.subject	Genetic algorithm	en
dc.subject	Input-output analysis	en
dc.subject	Logistic regression	en
dc.subject	Maximization of expected utility	en
dc.subject	Monte carlo methods	en
dc.subject	Prediction	en
dc.subject	Quality of health care	en
dc.subject	Sickness at hospital admission	en
dc.subject	Simulated annealing	en
dc.subject	Tabu search	en
dc.subject	Variable selection	en
dc.subject.classification	Statistics & Probability	en
dc.subject.other	PROSPECTIVE PAYMENT SYSTEM	en
dc.subject.other	HOSPITAL MORTALITY DATA	en
dc.subject.other	MEDICARE PATIENTS	en
dc.subject.other	QUALITY	en
dc.subject.other	CARE	en
dc.subject.other	PERFORMANCE	en
dc.subject.other	REGRESSION	en
dc.subject.other	ISSUES	en
dc.subject.other	CHOICE	en
dc.title	Comparing stochastic optimization methods for variable selection in binary outcome prediction, with application to health policy	en
heal.type	journalArticle	en
heal.identifier.primary	10.1198/016214508000001048	en
heal.identifier.secondary	http://dx.doi.org/10.1198/016214508000001048	en
heal.language	English	en
heal.publicationDate	2008	en
heal.abstract	Traditional variable-selection strategies in generalized linear models (GLMs) seek to optimize a measure of predictive accuracy without regard for the cost of data collection. When the purpose of such model building is the creation of predictive scales to be used in future studies with constrained budgets, the standard approach may not be optimal. We propose a Bayesian decision-theoretic framework for variable selection in binary-outcome GLMs where the budget for data collection is constrained and potential predictors may vary considerably in cost. The method is illustrated using data from a large study of quality of hospital care in the U.S. in the 1980s. Especially when the number of available predictors p is large, it is important to use an appropriate technique for optimization (e.g., in an application presented here where p = 83, the space over which we search has 283 = 1025 elements, which is too large to explore using brute force enumeration). Specifically, we investigate simulated annealing (SA), genetic algorithms (GAs), and the tabu search (TS) method used in operations research, and we develop a context-specific version of SA, improved simulated annealing (ISA), that performs better than its generic counterpart. When p was modest in our study, we found that GAs performed relatively poorly for all but the very best user-defined input configurations, generic SA did not perform well, and TS had excellent median performance and was much less sensitive to suboptimal choice of user-defined inputs. When p was large in our study, the best versions of GA and ISA outperformed TS and generic SA. Our results are presented in the context of health policy but can apply to other quality assessment settings with dichotomous outcomes as well. © 2008 American Statistical Association.	en
heal.publisher	AMER STATISTICAL ASSOC	en
heal.journalName	Journal of the American Statistical Association	en
dc.identifier.doi	10.1198/016214508000001048	en
dc.identifier.isi	ISI:000263008900008	en
dc.identifier.volume	103	en
dc.identifier.issue	484	en
dc.identifier.spage	1367	en
dc.identifier.epage	1381	en