Open Collections

Description

Quality of a process is often characterized by the functional relationship between a response and one or more predictors. Profile monitoring is for checking the stability of this relationship over time. In the literature, most existing control charts are for monitoring parametric profiles, and they assume that within-profile observations are independent of each other, which is often invalid. This talk presents some of our recent research on nonparametric profile monitoring when within-profile data are correlated. We will also briefly describe the problems of online image monitoring and dynamic disease screening that are closely related to profile monitoring.