國立虎尾科技大學 |

Variable selection and statistical learning for censored data.

紀錄類型:	書目-語言資料,手稿 : Monograph/item
正題名/作者:	Variable selection and statistical learning for censored data./
作者:	Liu, Xiaoxi.
面頁冊數:	1 online resource (137 pages)
附註:	Source: Dissertation Abstracts International, Volume: 76-01(E), Section: B.
Contained By:	Dissertation Abstracts International76-01B(E).
標題:	Biostatistics. -
電子資源:	click for full text (PQDT)
ISBN:	9781321142242

Variable selection and statistical learning for censored data.
Liu, Xiaoxi.

Variable selection and statistical learning for censored data. - 1 online resource (137 pages)

Source: Dissertation Abstracts International, Volume: 76-01(E), Section: B.

Thesis (Ph.D.)--The University of North Carolina at Chapel Hill, 2014.

Includes bibliographical references

This dissertation focuses on (1) developing an efficient variable selection method for a class of general transformation models; (2) developing a support vector based method for predicting failure times allowing the coarsening at random assumption for the censoring distribution; (3) developing a statistical learning method for predicting recurrent events.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2018

Mode of access: World Wide Web

ISBN: 9781321142242Subjects--Topical Terms:

783654
Biostatistics.
Index Terms--Genre/Form:

554714
Electronic books.

Variable selection and statistical learning for censored data.
LDR:03683ntm a2200361Ki 4500 001 918736
005 20181030085012.5
006 m o u
007 cr mn||||a|a||
008 190606s2014 xx obm 000 0 eng d
020 $a 9781321142242
035 $a (MiAaPQ)AAI3634028
035 $a (MiAaPQ)unc:14707
035 $a AAI3634028
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Liu, Xiaoxi. $3 1193138
245 1 0 $a Variable selection and statistical learning for censored data.
264 0 $c 2014
300 $a 1 online resource (137 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertation Abstracts International, Volume: 76-01(E), Section: B.
500 $a Adviser: Donglin Zeng.
502 $a Thesis (Ph.D.)--The University of North Carolina at Chapel Hill, 2014.
504 $a Includes bibliographical references
520 $a This dissertation focuses on (1) developing an efficient variable selection method for a class of general transformation models; (2) developing a support vector based method for predicting failure times allowing the coarsening at random assumption for the censoring distribution; (3) developing a statistical learning method for predicting recurrent events.
520 $a In the first topic, we propose a computationally simple method for variable selection in a general class of transformation models with right-censored survival data. The proposed algorithm reduces to maximizing a weighted partial likelihood function within an adaptive lasso framework. We establish the asymptotic properties for the proposed method, including selection consistency and semiparametric efficiency of parameter estimators. We conduct simulation studies to investigate the small-sample performance. We apply the method to data sets from a primary biliary cirrhosis study and the Atherosclerosis Risk in Communities (ARIC) Study, and demonstrate its superior prediction performance as compared to existing risk scores.
520 $a In the second topic, we develop a novel support vector hazard regression approach for predicting survival outcomes. Our method adapts support vector machines to predict dichotomous outcomes of the counting processes among subjects at risk, and allows censoring times to depend on covariates without modeling the censoring distribution. The formulation can be solved conveniently using any convex quadratic programming package. Theoretically, we show that the decision rule is equivalent to maximizing the discrimination power based on hazard functions, and establish the consistency and learning rate of the predicted risk. Numerical experiments demonstrate a superior performance of the proposed method to existing learning methods. Real data examples from a study of Huntington's disease and the ARIC Study are used to illustrate the proposed method.
520 $a In the third topic, we adapt support vector machines in the context of the counting process to handle time-varying covariates and predict recurrent events. We conduct extensive simulation studies to compare performances of the proposed method to the Andersen and Gill proportional intensity model for the prediction of multiple recurrences. The extension of theoretical properties is described. We illustrate the proposed method by analyzing the data set from a bladder cancer study.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2018
538 $a Mode of access: World Wide Web
650 4 $a Biostatistics. $3 783654
655 7 $a Electronic books. $2 local $3 554714
690 $a 0308
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a The University of North Carolina at Chapel Hill. $b Biostatistics. $3 1193116
773 0 $t Dissertation Abstracts International $g 76-01B(E).
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3634028 $z click for full text (PQDT)