Chemometrics and intelligent laboratory systems vol:75 issue:2 pages:127-136
Since MATLAB is very popular in industry and academia, and is frequently used by chemometricians, statisticians, chemists, and engineers, we introduce a MATLAB library of robust statistical methods. Those methods were developed because their classical alternatives produce unreliable results when the data set contains outlying observations. Our toolbox currently contains implementations of robust methods for location and scale estimation, covariance estimation (FAST-MCD), regression (FAST-LTS, MCD-regression), principal component analysis (RAPCA, ROBPCA), principal component regression (RPCR), partial least squares (RSIMPLS) and classification (RDA). Only a few of these methods will be highlighted in this paper. The toolbox also provides many graphical tools to detect and classify the outliers. The use of these features will be explained and demonstrated through the analysis of some real data sets. (C) 2004 Elsevier B.V. All rights reserved.