Phase I monitoring and change point estimation of autocorrelated poisson regression profiles

Performance evaluation of DEWMA3 in phase-II for capturing changes in simple linear profiles based on run rule mechanism

In Statistical Process Control, many techniques exist for monitoring the stability of a process over time. In this work, we study the relationship of the response variable with explanatory variables in the form of linear profiles for detecting changes in slope and intercept of the linear quality profiles. We used the transformation of explanatory variables approach used for make the regression estimates independent of each other to have zero average. A comparative study of three phase-II methods using DEWMA statistics in monitoring and capturing undesirable deviations in the slope, intercept, and variability is also studied by applying different proposed run rules schemes i.e., R_1/1, R_2/3, R_3/3. Monte Carlo simulations were carried out on R-Software for finding the results of proposed schemes by taking various levels of shifts for intercept, slope, and standard deviation in identifying the false alarm rate of a process. The simulation results based on the average run length criterion show that the proposed run rule schemes improve the detection ability of the control structure. Among all the proposed schemes R_2/3 is found to be the best one because of its quick detection ability of false alarm rate. The proposed scheme also shows superiority in comparison to other schemes. The simulation results are further justified with a real data application.