Abstract
This article is concerned with the weighted least squares approach to the analysis of digital information system in industrial experiments. The odds ratios between input and output play an important role in the four-fold tables for orthogonal experiments with categorical data. Diagnostic tools are also developed to examine outliers and influence of individual experiments for assessing a model's fit and stability. To illustrate how the methods proposed in this article can be used, we analyze an industrial experiment data, and insights into the data are obtained.
