Dictionary:Principal component analysis (PCA)
1. A procedure that transforms a number of possibly correlated variables into a smaller number of uncorrelated variables (the principal components). It ranks the principal components according to the amount of the data variability for which each accounts, so that components whose effects are only minor can be ignored. Principal component analysis is generally used to identify the meaningful variables and reduce the dimensionality of the data set. In an eigenanalysis, the first principal component is in the same direction as the eigenvector associated with the largest eigenvalue, and so on to other components in descending order of importance. PCA is often done by singular-value decomposition (q.v.). 2. Where images correlate because they contain portions of the same information, the separation of the information into orthogonal images. For example, much of the information on different Landsat bands correlate and PCA separates the information into uncorrelated images.