Exploratory Factor Analysis
Abstrak
PCA and SVD are considered simple forms of exploratory factor analysis. The term ‘factor analysis’ is a bit confusing and you will find a variety of definitions out there–some people assert that PCA is not factor analysis, and others might use PCA but call it factor analysis. PCA involves a complete redescription of the covariance or correlation matrix along a set of independent dimensions, until it completely redescribes the data, and then you pick off the most important components and ignore the rest. EFA is typically attempting to do the same thing, but trying to maximize the variance accounted for by a fixed number of ‘latent’ factors, so that we have the model + error. Furthermore, most of the time factor analysis lets us pick orthogonal dimensions based on criteria other than maximum variance, and permits factors that are not orthogonal, allowing for latent factors that are somewhat related.
Topik & Kata Kunci
Penulis (2)
J. Cappelleri
Robert A. Gerber
Akses Cepat
- Tahun Terbit
- 2018
- Bahasa
- en
- Total Sitasi
- 1808×
- Sumber Database
- Semantic Scholar
- DOI
- 10.1201/9781351110273-130001078
- Akses
- Open Access ✓