Outliers and robust procedures in some chemometric applications

Environmental chemists generate and analyze large amounts of data. They often need to identify experimental conditions and results which look suspicious and are significantly different from the rest of the data. These observations are known as outliers. The robust procedure based upon a recently proposed influence function, called the PROP function is resistant to masking and swamping. The procedure works effectively in correctly identifying multiple univariate and multivariate outliers, and also in obtaining reliable estimates of population parameters in several areas of interest including multiple linear regression and principal component analyses. The control-chart-type quantile-quantile (Q-Q) plot of Mahalanobis distances provides a formal graphical multiple outlier identification procedure. Also the control-chart-type scatter plot of the robustified square root leverage distances versus the residuals identifies all regression outliers and distinguishes between bad and good leverage points.