Robust statistics in data analysis -- A review Basic concepts

摘要

Presence of outliers in chemical data affects all least squares models, which are extensively used in chemometrics for data exploration and modeling. Therefore, more and more attention is paid to the so-called robust models and robust statistics that aim to construct models and estimates describing well data majority. Moreover, construction of robust models allows identifying outlying observations. The outliers identification is not only essential for a proper modeling but also for understanding the reasons for unique character of the outlying sample. In this paper some basic concepts of robust techniques are presented and their usefulness in chemometric data analysis is stressed.