Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

An algorithm for sparse factor analysis with common factors and/or specific factors dissociated from errors

An algorithm for sparse factor analysis with common factors and/or specific factors dissociated... Two new approaches to factor analysis (FA) were presented in this century. One of them is developing the FA procedure underlain by a comprehensive FA (CompFA) model. In this model, a multivariate observation is decomposed into common factor, specific factor, and errors parts, with the three parts assumed to be uncorrelated mutually. The other approach is modifying FA so as to provide a sparse loading matrix. Such a modification is called sparse FA. The two approaches are combined in this note: its purpose is to present an algorithm for the sparse FA procedure underlain by the CompFA model, with the algorithm based on an alternating direction method of multipliers. The algorithm also covers the variants of the procedure, in which the model assumption is relaxed so that either common or specific factor part may be correlated with the error part. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Behaviormetrika Springer Journals

An algorithm for sparse factor analysis with common factors and/or specific factors dissociated from errors

Adachi, Kohei
Behaviormetrika , Volume OnlineFirst – Feb 1, 2023
Download PDF
12 pages

Loading next page...
 
/lp/springer-journals/an-algorithm-for-sparse-factor-analysis-with-common-factors-and-or-YZepHmG1Zw
Publisher
Springer Journals
Copyright
Copyright © The Behaviormetric Society 2023
ISSN
0385-7417
eISSN
1349-6964
DOI
10.1007/s41237-023-00195-1
Publisher site
See Article on Publisher Site

Abstract

Two new approaches to factor analysis (FA) were presented in this century. One of them is developing the FA procedure underlain by a comprehensive FA (CompFA) model. In this model, a multivariate observation is decomposed into common factor, specific factor, and errors parts, with the three parts assumed to be uncorrelated mutually. The other approach is modifying FA so as to provide a sparse loading matrix. Such a modification is called sparse FA. The two approaches are combined in this note: its purpose is to present an algorithm for the sparse FA procedure underlain by the CompFA model, with the algorithm based on an alternating direction method of multipliers. The algorithm also covers the variants of the procedure, in which the model assumption is relaxed so that either common or specific factor part may be correlated with the error part.

Journal

Behaviormetrika – Springer Journals

Published: Feb 1, 2023

Keywords: Sparse factor analysis; Comprehensive factor analysis model; Completely decomposed factor analysis; Alternating direction method of multipliers; Unpenalized approach; Extended orthogonal Procrustes rotation

References

  • Factor analysis procedures revisited from the comprehensive model with the unique factors decomposed into specific factors and errors
    Adachi, K
  • Sparse orthogonal factor analysis
    Adachi, K; Trendafilov, NT; Carpita, E; Brentari, E; Qannari, EM
  • Sparse principal component analysis subject to prespecified cardinality of loadings
    Adachi, K; Trendafilov, NT
  • Distributed optimization and statistical learning via the alternating direction method of multipliers
    Boyd, S; Parikh, N; Chu, B; Peleato, B; Eckstein, J
  • The development of markers for the big-five factor structure
    Goldberg, LR
  • Sparse and simple structure estimation via prenet penalization
    Hirose, K; Terada, Y
  • Estimation of an oblique structure via penalized likelihood factor analysis
    Hirose, K; Yamamoto, M
  • Sparse estimation via nonconcave penalized likelihood in factor analysis model
    Hirose, K; Yamamoto, M
  • Statistical inference of minimum rank factor analysis
    Shapiro, A; ten Berge, JMF
  • A new method for simultaneous estimation of the factor model parameters, factor scores, and unique parts
    Stegeman, A
  • Orthogonal rotations to maximal agreement for two or more matrices of different column orders
    ten Berge, JMF; Knol, DL
  • Factor analysis
    Yanai, H; Ichikawa, M; Rao, CR; Sinharay, S