MISSING VALUES IN PRINCIPAL COMPONENT ANALYSIS

Calculation schemes for principal component analysis are considered fo r the case when some matrix elements are missing. Iterative solutions are proposed-either a set of multilinear regression problems or as sin gular-value decomposition problems with iterative imputation of missin g values. If mean values are subtracted from the data matrix, they sho uld also be included in the iteration scheme. Test calculations using Matlab show that the regression approach is somewhat faster than the i mputation approach. The results with a substantial amount of missing d ata are different and superior to those obtained with the naive NIPALS algorithm in common use in chemometrics. (C) 1998 Elsevier Science B. V. All rights reserved.