Multivariate EWMA control chart with highly asymmetric gamma distributions. Issue 2 (4th March 2018)

• Record Type: Journal Article
• Authors: Flury, María I.; Quaglino, Marta B.
• Is Part Of: Quality technology & quantitative management. Volume 15:Issue 2(2018)
• Page Start: 230
• Page End: 252
• Publication Date: 2018-03-04
• Subjects: Average run length (ARL) -- multivariate statistical process control (MSPC) -- probabilistic control limits -- T2 chart; Quality control -- Periodicals; Quality control -- Statistical methods -- Periodicals; Industrial management -- Periodicals; Industrial management; Management -- Research -- Methodology -- Periodicals; Qualitative research -- Periodicals; Management; Quality control; Quality control -- Statistical methods; Periodicals; 658.00721
• DOI: 10.1080/16843703.2016.1208937
• Languages: English
• ISSNs: 1684-3703

Abstract

Abstract: In this paper, we investigated the performance of the multivariate exponentially weighted moving average (MEWMA) control chart under highly asymmetric gamma distributions. Simulation procedures were used to obtain probabilistic control limits of the chart using a specific underlying distribution with known parameters (Phase II). The performance of the chart was evaluated under three different mean vector shifts, and the best values for the smoothing constant were identified. Results were compared with the existing proposal of using normal limits. Our results suggested that for three or more variables, probabilistic control limits lead to a better performance of the chart. We extended the analysis to the T 2 chart and compared the efficiency of both strategies. MEWMA chart performance was better than T 2 chart in all the cases analysed. A real application with Phase I reference data is described. The parameters of the marginal gamma distributions were estimated and a specific algorithm was then employed to determine the appropriate values of the smoothing constant and upper control limit of the MEWMA chart (Phase II). The relevance of this method lies in the fact that it can be adapted to any practical situation in which gamma model is suitable.