A data analysis strategy for quality engineering experiments

This paper deals with techniques for analyzing data from quality engineering experiments for optimizing a process with fixed target. We propose a structured data-analytic approach with three phases of analysis: an exploratory phase, a modeling phase, and an optimization phase. We emphasize the use of graphical methods in all three phases to guide the analysis and facilitate interpretation of results. We discuss the role of data transformations and the relationship between an analysis of transformed data and Taguchi's signal-to-noise ratio analysis. Our strategy is presented in a form that can be easily followed by users, and the various steps are illustrated by an example.