|
Activity Number:
|
504
- Model/Variable Selection
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 2, 2017 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
|
Abstract #322533
|
|
|
Title:
|
Model Selection in Latent Class Analysis
|
|
Author(s):
|
Zhaoyin Zhu* and Yongzhao Shao
|
|
Companies:
|
NYU and New York University School of Medicine
|
|
Keywords:
|
Latent class analysis ;
Variable selection ;
Clustering ;
EM algorithm
|
|
Abstract:
|
As one of the most widely used model-based clustering methods for categorical data, latent class analysis (LCA) is becoming an increasingly popular tool to identify and characterize the unmeasured latent classes underlying data in a broad range of applied fields. After including covariates into the basic latent class model to predict individuals' latent class membership, how to select the number of latent classes and identify significant risk factors becomes a challenging task. The aim of this paper is to develop a method to select both the number of classes and the related variable via regularized approach. The performance of our model selection method is illustrated via simulation studies. The theoretical foundations for making large sample inference are established as well.
|
Authors who are presenting talks have a * after their name.
