Parameter Estimation in Factor Analysis: Maximum Likelihood versus Principal Component

被引：14

作者：

Kassim, Suraiya ^{[1
]}

Hasan, Husna ^{[1
]}

Ismon, Aisyah Mohd ^{[1
]}

Asri, Fahimah Muhammad ^{[1
]}

机构：

[1] Univ Sains Malaysia, Sch Math Sci, Minden 11800, Penang, Malaysia

来源：

PROCEEDINGS OF THE 20TH NATIONAL SYMPOSIUM ON MATHEMATICAL SCIENCES (SKSM20): RESEARCH IN MATHEMATICAL SCIENCES: A CATALYST FOR CREATIVITY AND INNOVATION, PTS A AND B | 2013年 / 1522卷

关键词：

factor analysis; principal component estimator; maximum likelihood estimator; COMMON FACTOR-ANALYSIS;

D O I：

10.1063/1.4801279

中图分类号：

O29 [应用数学];

学科分类号：

070104 ;

摘要：

Factor analysis (FA) is a multivariate statistical technique to uncover latent constructs of observed variables while principal component analysis (PCA) is a technique to reduce the number of variables in large data sets. While both are different techniques with different objectives, they often produced the same factor solution. In this paper, the multivariate statistical theory behind PCA and FA is discussed to highlight their differences and similarities. In particular, the factor extraction technique of PCA is compared to the extraction procedures of principal axis factoring (PAF) and maximum likelihood estimation (MLE) in factor analysis. Results from applying the procedures on a published data set will be discussed and the appropriate use of each will be suggested.

引用

页码：1293 / 1299

页数：7