This package provides several functions that mainly use EM algorithm to fit probabilistic PCA and Factor analysis models.
PPCA is probabilistic counterpart of PCA model. PPCA has the advantage that it can be further extended to more advanced model, such as mixture of PPCA, Bayeisan PPCA or model dealing with missing data, etc. However, this package mainly served a research and teaching purpose for people to understand the model. The code is succinct so that it is easy to read and learn.
This package is now a part of the PRML toolbox (http://cn.mathworks.com/help/stats/ppca.html).
Mo Chen (2019). Probabilistic PCA and Factor Analysis (https://www.mathworks.com/matlabcentral/fileexchange/55883-probabilistic-pca-and-factor-analysis), MATLAB Central File Exchange. Retrieved .
1.0.0.0 | update description |
|
1.0.0.0 | update description |
Inspired by: Pattern Recognition and Machine Learning Toolbox
Create scripts with code, output, and formatted text in a single executable document.
Kris Villez (view profile)
Just tested the fa.m function. Works well but had to decrease the tolerance to 1e-8 to make have accuracy comparable to 'factoran' on some challenging data sets.