Top Page | Upper Page | Contents | About This Site | JAPANESE

Principal Component Analysis

Principal Component Analysis (PCA) changes ,any X variables into a few new variables. Tne new variable is called "principal component". The new variables tell us the main information of the data set.

New variables have orthogonal relationship. So, if we use the variables as the X of multi-regression analysis , the analysis is easier.

For Category Data, Broadly defined Quantification theory 3 is available as PCA.

Process of PCA

New variable is made to maximize the variance of it. Maximization is done using Lagrange's method of undetermined multipliers.

The analysist have to consider the meanings of each new variable.


Example of R is in the page, Principal component analysis by R .

Principal Component Regression Analysis

Principal Component MT

Analysis Using Intermediate Layer

Factor Analysis

NEXT Canonical Correlation Analysis