Search

Continuous process improvement is an important issue for industry processes. In order to achieve this prominent objective, recent years have witnessed considerable interest in the prospects for process enhancement to be realized by integrated usc of statistical process control (SPC) and engineering process control (EPC). Most of the studies assumed that the assignable causes or disturbances could be identified and removed right away as long as an out-of-control signal is triggered by employing the SPC mechanism of the integrated system. However, for a practical consideration, it is not straightforward to identify, isolate, remove, and correct the causes of disturbance. This study aims to develop techniques which effectively integrate SPC and EPC so that the system is able to distinguish two well-known classes of disturbances - step change and linear - from one another. The use and the potential advantages of the proposed techniques are illustrated through simulation studies.

The sales promotion data resulting from multiple marketing strategies are usually autocorrelated. Consequently, the characteristics of those data sets can be analyzed using time-series and/or intervention analysis. Traditional time-series intervention analysis focuses on the effects of single or few interventions, and forecasts may be obtained as long as the future interventions can be assured. This study is different from traditional approaches, and considers the cases in which multiple interventions and the uncertainty of future interventions exist in the system. In addition, this study utilizes a set of real sales promotion data to demonstrate the effectiveness of the proposed approach.