This book is Work in Progress. I appreciate your feedback to make the book better.

8.1 Components

Principal component analysis

Principal component analysis summarizes many observed variables with a smaller number of weighted combinations. Components are constructed directly from the observed variables. They can be useful for compression, visualization, and prediction, but they do not by themselves posit an unobserved construct that generates the indicators.

This distinction becomes important later: components summarize observed variation; factor models explain shared variation through latent variables.