FPCASCORE Procedure

The FPCASCORE procedure applies a previously trained functional principal component analysis (FPCA) model to new functional data in order to compute functional principal component scores. You would typically use this procedure to score or project new curves onto the principal components that are derived from a training data set by the FPCA procedure.

PROC FPCASCORE requires a saved analytic store (created by the SAVESTATE statement in PROC FPCA), which contains the estimated mean function and eigenfunctions from the training data. The procedure computes projection scores for the input data by using these stored components.

This scoring capability is useful in tasks such as anomaly detection, functional classification, and forecasting, or in any application that benefits from dimensionality reduction of functional data.