Discussion of recent advances in process monitoring: Nonparametric and variable-selection methods for phase i and phase II

Abstract

Emmanuel Yashchin, IBM, congratulates Dr. Capizzi on her paper titled 'Recent Advances in Process Monitoring: Nonparametric and Variable-Selection Methods for Phase I and Phase II'. The author finds that in the massive data environment statistical process monitoring (SPM) can benefit greatly from more analytically advanced procedures. The general approach to the SPM was based on use of Cumulative Sum (CUSUM)-Shewhart technique proved to be extensible to general data structures and well received by technical personnel. He also found that in the massive data environment there tend to be too many anomalies that are of no interest to users, so detection schemes based on statistical significance, like 3-sigma Shewhart schemes, tend to be of rather limited use. Of special value is the ability of an SPM system to guide the users toward understanding of the underlying causes. The paper focuses on the Phase 1 data that are expected to conform to an independent and identically distributed pattern. Of special value are tests, such as RS/P, which are essential prior to derivation of control limits for any type of detection schemes.