Automatic monitoring and analysis of engineering experiments

Yield and process improvement experiments generate enormous amounts of data that need to be analyzed. The sheer volume of data tends to overwhelm even the most enthusiastic engineer. A "hands-off" analysis routine that did all the number crunching for them was needed. This paper describes a program that runs in the background, analyzing experimental results using a stepwise regression technique that tests for multivariate effects and a nonparametric analysis of variance. It produces and stores an array of charts and summaries in a standard format for review by any one in the engineering staff.