國立高雄大學圖資館 |

登入

回首頁

Model-Based Clustering of Longitudinal Data in High Dimensions.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Model-Based Clustering of Longitudinal Data in High Dimensions.
作者:	Yang, Luoying.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, 2021
面頁冊數:	141 p.
附註:	Source: Dissertations Abstracts International, Volume: 83-05, Section: B.
附註:	Advisor: Wu, Tongtong.
Contained By:	Dissertations Abstracts International83-05B.
標題:	Statistics.
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28719737
ISBN:	9798460480647

Model-Based Clustering of Longitudinal Data in High Dimensions.
Yang, Luoying.

Model-Based Clustering of Longitudinal Data in High Dimensions. - Ann Arbor : ProQuest Dissertations & Theses, 2021 - 141 p.

Source: Dissertations Abstracts International, Volume: 83-05, Section: B.

Thesis (Ph.D.)--University of Rochester, 2021.

This item must not be sold to any third party vendors.

We propose a model-based clustering method for high-dimensional longitudinal data via regularization. This study was motivated by the Trial of Activity in Adolescent Girls (TAAG), which aimed to examine multi-level factors related to the change of physical activities by following up a cohort of 783 girls over 10 years from adolescence to early adulthood. Our goal is to identify the intrinsic grouping of subjects with similar patterns of physical activity trajectories and the most relevant predictors among over 800 candidate variables within groups. The previous analyses could only conduct clustering and variable selection in two steps, while our new method can perform the tasks simultaneously. Within each cluster, a linear mixed-eﬀects model is fitted with a doubly penalized likelihood to induce sparsity for efficient parameter identification and estimation. The large-sample properties are established, allowing the dimension of both ﬁxed and random eﬀects to increase at an exponential rate of the sample size, with a general class of penalty functions. Assuming subjects are drawn from a Gaussian mixture distribution, model eﬀects and cluster labels are estimated via a coordinate descent algorithm nested inside the Expectation-Maximization algorithm. Bayesian Information Criterion is used to determine the optimal number of clusters and values of tuning parameters. Our numerical studies show that the new method has satisfactory performance and is able to accommodate complex data with multi-level and/or longitudinal effects.

ISBN: 9798460480647Subjects--Topical Terms:

182057
Statistics.
Subjects--Index Terms:

Exponentially growing number of variables

Model-Based Clustering of Longitudinal Data in High Dimensions.
LDR:02747nmm a2200361 4500 001 616470
005 20220513114345.5
008 220920s2021 ||||||||||||||||| ||eng d
020 $a 9798460480647
035 $a (MiAaPQ)AAI28719737
035 $a AAI28719737
040 $a MiAaPQ $c MiAaPQ
100 1 $a Yang, Luoying. $0 (orcid)0000-0003-1653-8350 $3 915808
245 1 0 $a Model-Based Clustering of Longitudinal Data in High Dimensions.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2021
300 $a 141 p.
500 $a Source: Dissertations Abstracts International, Volume: 83-05, Section: B.
500 $a Advisor: Wu, Tongtong.
502 $a Thesis (Ph.D.)--University of Rochester, 2021.
506 $a This item must not be sold to any third party vendors.
520 $a We propose a model-based clustering method for high-dimensional longitudinal data via regularization. This study was motivated by the Trial of Activity in Adolescent Girls (TAAG), which aimed to examine multi-level factors related to the change of physical activities by following up a cohort of 783 girls over 10 years from adolescence to early adulthood. Our goal is to identify the intrinsic grouping of subjects with similar patterns of physical activity trajectories and the most relevant predictors among over 800 candidate variables within groups. The previous analyses could only conduct clustering and variable selection in two steps, while our new method can perform the tasks simultaneously. Within each cluster, a linear mixed-eﬀects model is fitted with a doubly penalized likelihood to induce sparsity for efficient parameter identification and estimation. The large-sample properties are established, allowing the dimension of both ﬁxed and random eﬀects to increase at an exponential rate of the sample size, with a general class of penalty functions. Assuming subjects are drawn from a Gaussian mixture distribution, model eﬀects and cluster labels are estimated via a coordinate descent algorithm nested inside the Expectation-Maximization algorithm. Bayesian Information Criterion is used to determine the optimal number of clusters and values of tuning parameters. Our numerical studies show that the new method has satisfactory performance and is able to accommodate complex data with multi-level and/or longitudinal effects.
590 $a School code: 0188.
650 4 $a Statistics. $3 182057
650 4 $a Biostatistics. $3 339693
653 $a Exponentially growing number of variables
653 $a Linear mixed-effects model
653 $a Longitudinal analysis
653 $a Nonconcave penalty functions
653 $a Simultaneous effects selection
690 $a 0463
690 $a 0308
710 2 $a University of Rochester. $b School of Medicine and Dentistry. $3 915809
773 0 $t Dissertations Abstracts International $g 83-05B.
790 $a 0188
791 $a Ph.D.
792 $a 2021
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28719737