Cluster analysis of longitudinal profiles with subgroups

Xiaolu Zhu, Annie Qu

Research output: Contribution to journalArticle

Abstract

In this paper, we cluster profiles of longitudinal data using a penalized regression method. Specifically, we allow heterogeneous variation of longitudinal patterns for each subject, and utilize a pairwise-grouping penalization on coefficients of the nonparametric B-spline models to form subgroups. Consequently, we identify clusters based on different patterns of the predicted longitudinal curves. One advantage of the proposed method is that there is no need to pre-specify the number of clusters; instead the number of clusters is selected automatically through a model selection criterion. Our method is also applicable for unbalanced data where different subjects could have measurements at different time points. To implement the proposed method, we develop an alternating direction method of multipliers (ADMM) algorithm which has the desirable convergence property. In theory, we establish the consistency properties for approximated nonparametric function estimation and subgrouping memberships. In addition, we show that our method outperforms the existing competitive approaches in our simulation studies and real data example.

Original languageEnglish (US)
Pages (from-to)171-193
Number of pages23
JournalElectronic Journal of Statistics
Volume12
Issue number1
DOIs
StatePublished - Jan 1 2018

Fingerprint

Cluster Analysis
Subgroup
Number of Clusters
Method of multipliers
Unbalanced Data
Penalized Regression
Alternating Direction Method
Model Selection Criteria
Function Estimation
Penalization
Nonparametric Estimation
Longitudinal Data
B-spline
Grouping
Convergence Properties
Pairwise
Profile
Cluster analysis
Simulation Study
Curve

Keywords

  • ADMM
  • Longitudinal data
  • Minimax concave penalty
  • Model selection
  • Nonparametric spline method

ASJC Scopus subject areas

  • Statistics and Probability

Cite this

Cluster analysis of longitudinal profiles with subgroups. / Zhu, Xiaolu; Qu, Annie.

In: Electronic Journal of Statistics, Vol. 12, No. 1, 01.01.2018, p. 171-193.

Research output: Contribution to journalArticle

@article{88532a58b7814048860179d631550705,
title = "Cluster analysis of longitudinal profiles with subgroups",
abstract = "In this paper, we cluster profiles of longitudinal data using a penalized regression method. Specifically, we allow heterogeneous variation of longitudinal patterns for each subject, and utilize a pairwise-grouping penalization on coefficients of the nonparametric B-spline models to form subgroups. Consequently, we identify clusters based on different patterns of the predicted longitudinal curves. One advantage of the proposed method is that there is no need to pre-specify the number of clusters; instead the number of clusters is selected automatically through a model selection criterion. Our method is also applicable for unbalanced data where different subjects could have measurements at different time points. To implement the proposed method, we develop an alternating direction method of multipliers (ADMM) algorithm which has the desirable convergence property. In theory, we establish the consistency properties for approximated nonparametric function estimation and subgrouping memberships. In addition, we show that our method outperforms the existing competitive approaches in our simulation studies and real data example.",
keywords = "ADMM, Longitudinal data, Minimax concave penalty, Model selection, Nonparametric spline method",
author = "Xiaolu Zhu and Annie Qu",
year = "2018",
month = "1",
day = "1",
doi = "10.1214/17-EJS1389",
language = "English (US)",
volume = "12",
pages = "171--193",
journal = "Electronic Journal of Statistics",
issn = "1935-7524",
publisher = "Institute of Mathematical Statistics",
number = "1",

}

TY - JOUR

T1 - Cluster analysis of longitudinal profiles with subgroups

AU - Zhu, Xiaolu

AU - Qu, Annie

PY - 2018/1/1

Y1 - 2018/1/1

N2 - In this paper, we cluster profiles of longitudinal data using a penalized regression method. Specifically, we allow heterogeneous variation of longitudinal patterns for each subject, and utilize a pairwise-grouping penalization on coefficients of the nonparametric B-spline models to form subgroups. Consequently, we identify clusters based on different patterns of the predicted longitudinal curves. One advantage of the proposed method is that there is no need to pre-specify the number of clusters; instead the number of clusters is selected automatically through a model selection criterion. Our method is also applicable for unbalanced data where different subjects could have measurements at different time points. To implement the proposed method, we develop an alternating direction method of multipliers (ADMM) algorithm which has the desirable convergence property. In theory, we establish the consistency properties for approximated nonparametric function estimation and subgrouping memberships. In addition, we show that our method outperforms the existing competitive approaches in our simulation studies and real data example.

AB - In this paper, we cluster profiles of longitudinal data using a penalized regression method. Specifically, we allow heterogeneous variation of longitudinal patterns for each subject, and utilize a pairwise-grouping penalization on coefficients of the nonparametric B-spline models to form subgroups. Consequently, we identify clusters based on different patterns of the predicted longitudinal curves. One advantage of the proposed method is that there is no need to pre-specify the number of clusters; instead the number of clusters is selected automatically through a model selection criterion. Our method is also applicable for unbalanced data where different subjects could have measurements at different time points. To implement the proposed method, we develop an alternating direction method of multipliers (ADMM) algorithm which has the desirable convergence property. In theory, we establish the consistency properties for approximated nonparametric function estimation and subgrouping memberships. In addition, we show that our method outperforms the existing competitive approaches in our simulation studies and real data example.

KW - ADMM

KW - Longitudinal data

KW - Minimax concave penalty

KW - Model selection

KW - Nonparametric spline method

UR - http://www.scopus.com/inward/record.url?scp=85041376915&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85041376915&partnerID=8YFLogxK

U2 - 10.1214/17-EJS1389

DO - 10.1214/17-EJS1389

M3 - Article

AN - SCOPUS:85041376915

VL - 12

SP - 171

EP - 193

JO - Electronic Journal of Statistics

JF - Electronic Journal of Statistics

SN - 1935-7524

IS - 1

ER -