MULTISTAGE PROCESS MONITORING USING ROBUST CAUSE SELECTING CONTROL CHART

Cause selecting charts (CSCs) are believed to be invaluable in multistage processes where an output variable of interest is impacted by one or more inputs. To establish the relationship between input and output variables, a standard procedure is to use historical data, which is often prone to hold outliers. The presence of outliers tends to decrease the effectiveness of monitoring procedures because the control limits become stretched. To dampen the negative repercussions of outliers, robust regression has been implemented instead of ordinary least squares method. A simulation study is included to investigate and compare the average run length (ARL) of robust and non-robust schemes. The results disclose that using robust procedure is more efficacious for detecting different types of shift when outliers exist.