BOX-COX TRANSFORMATIONS AND ROBUST CONTROL CHARTS IN SPC

Abstract

In this paper we consider some asymetric models to describe the data process. We assess that, in general, it is possible to find an adequate Box-Cox transformation which enables us to deal with the transformed data as approximately normal. To monitor the original and the transformed data we implement the traditional control charts for normal data as well as robust control charts, based on the total median and on the total range statistics. All these charts are compared in terms of efficiency and robustness.