Generalized linear model-based control charts for discrete semiconductor process data

In previous simple cases generalized linear model (GLM)-based control charts have been shown to be very effective in detecting shifts in multivariate counts when input variables are measurable. This paper studies the effectiveness of GLM-based control charts on more complicated data sets with multiple inputs and outputs whose relationships are varied. Results show that the GLM-based charts were most effective in detecting changes in the means of overdispersed counts (when compared with counts with normal dispersion). The GLM-based charts were more effective than multiple C charts in detecting changes in the means of counts when multiple complicated relationships exist.