The performance of robust multivariate Ewma control charts
Multivariate Exponential Weighted Moving Average (MEWMA) control chart is a popular statistical tool for monitoring multivariate process over time. However, this chart is sensitive to the presence of outliers arising from the use of classical mean vector and covariance matrix in estimating the MEWMA...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Academic Research Publishing Group
2018
|
| Subjects: | |
| Online Access: | http://repo.uum.edu.my/25768/1/TJSSR%206%202018%2052%2058.pdf http://repo.uum.edu.my/25768/ http://doi.org/10.32861/jssr.spi6.52.58 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Multivariate Exponential Weighted Moving Average (MEWMA) control chart is a popular statistical tool for monitoring multivariate process over time. However, this chart is sensitive to the presence of outliers arising from the use of classical mean vector and covariance matrix in estimating the MEWMA statistic. These classical
estimators are known to be sensitive to the outliers. To address this problem, robust MEWMA control charts based on modified one-step M-estimator (MOM) and Winsorized modified one-step M-estimator (WM) are proposed. Their performance is then compared with the standard MEWMA control chart in various situations. The findings
revealed that the proposed robust MEWMA control charts are more effective in controlling false alarm rates especially for large sample sizes and high percentage of outliers |
|---|