Using Sequential CUSUM Permutation Test to Identify Process Shifts
Nelson Lee Afanador, Senior Project Statistician; Hesham M. Fahmy, PhD, Scientific Fellow, Merck & Co.
A novel changepoint detection procedure based on a sequential CUSUM permutation test is presented. The new procedure provides process engineers with useful information in their search for special causes of variation, and could significantly reduce the cost and time of the root cause identification process. The new procedure was written in R, and a JMP GUI was built to integrate JMP with R to make the new tool more practical to use for engineers and scientists. In this presentation, we review the new approach of using sequential permutation tests for the purpose of objectively identifying change points in a manufacturing process.