ABSTRACT
This chapter is dedicated to functional principal components analysis (FPCA) and various extensions to non-Gaussian and sparsely sampled functional data and provides the R software for conducting these types of analyses. The function refund::fpca.face is used to implement FPCA for dense functional data with or without missing observations. Methods are extended to non-Gaussian functional data implemented in the fastGFPCA R package and sparse functional data implemented in the face::face.sparse R package. Methods are illustrated on the NHANES and CONTENT data sets. Examples of cases when PCA does not work well are also highlighted.
