The Performance of Bootstrap Confidence Interval of Robust Process Capability Index in Electronics Production Quality Control

Hanissah, Mohamad @ Sulaiman and Yosza, Dasril and Farah Shahnaz, Feroz and Norazlina, Abd Razak and Wong, Yan Chiew (2010) The Performance of Bootstrap Confidence Interval of Robust Process Capability Index in Electronics Production Quality Control. Project Report. UTeM, Melaka, Malaysia. (Submitted)

PDF (Full Text) The_Performance_Of_Bootstrap_Confidence_Interval_Of_Robust_Process_Capability_Index_In_Electronics_Production_Quality_Control_-_Hanissah_Binti_Mohamad_@_Sulaiman.pdf - Submitted Version Restricted to Registered users only Download (3MB)

Abstract

THE PERFORMANCE OF BOOTSTRAP CONFIDENCE INTERVAL OF ROBUST PROCESS CAPABILITY INDEX IN ELECTRONICS PRODUCTION QUALITY CONTROL A process capability index is numerical summary that compares the behavior of a product or process characteristic to engineering specifications. A Cpk index is used to measure whether a production process i capable of producing items that satisfy a customer requirements(i.e specification limits). The computation of the Cpk index is based on the sample mean, x and sample standard deviation, s which are known to be very sensitive to the presence of outliers. As an alternative, we may turn to the robust location and scale estimate based on a robust MM estimates which are less affected by outliers. A major step toward the correct understanding and interpretation of Cpk index is by constructing ifs confidence interval. The construction of such intervals assume that the measurement process having a normal distribution. However, many process are not normal and have a fat-tai led distribution which are prone to produce outliers. An alternative approach is to use bootstrap method such as the Percentile (P) and Bias-Corrected and Acceleration (Bca) for calculating approximates confidence intervals of Cpk index. It is computer intensive based method that can be utilized without relying any assumption on the underlying distribution. The results of the studies reveal that the Bca method seems to perform better than the Percentile method for both normal and skewed process. The performance of the Cpk-MM estimates were investigated for further by comparing the bootstrap confidence interval for Cpk index MM estimates and the well-known classical Cpk estimates. Based on simulation studies, show that the MM estimates produced more reliable confidence interval compared to the classical Cpk estimates.

Actions (login required)

View Item