Lower confidence bound for process-yield index S_pk with autocorrelated process data

In this paper, the lower confidence bound for the process-yield index S_pk is developed for autocorrelated process data. A simulation study is conducted to assess the performance of the proposed index and two existing indices C_p and C_pk for various combinations of sample size, autoregressive parameter, and true index value. The simulation results confirm that the estimated S_pk performs better than the other two indices C_p and C_pk regarding bias and standard deviation. Additionally, the coverage rates of Ŝ_pk in most cases are greater than a 95% lower limit of the stated nominal. Two real examples are used to demonstrate the application of the proposed approach.