CMStatistics 2017: Start Registration
View Submission - CMStatistics
B1760
Title: Monitoring occupational exposure data with joint control charts Authors:  M Rosario Ramos - FCiencias.ID (Portugal) [presenting]
Elisabete Carolino - ESTeSL-IPL (Portugal)
Carla Viegas - ESTeSL-IPL and Research Group Environment and Health of ESTeSL (Portugal)
Susana Viegas - ESTeSL-IPL and Research Group Environment and Health of ESTeSL (Portugal)
Abstract: Classical control systems using the Shewhart charts are often difficult to manage in real time and conditions of each environment. The use of univariate and independent control charts when measurements are relative to simultaneous observations, possibly correlated, distorts multivariate vector monitoring through type I error and the probability of being within the control limits will be not adjusted to the real value. This distortion of the control procedure increases with the number of variables to be controlled. Motivated by the need to promote a reliable monitoring system of the concentration of particles with adverse health effects, some alternatives are explored. A multivariate approach of control charts is used for the first time in the monitoring of the particulate matter in an animal feed production industry. Data are measurements on particles providing information about the concentration for particles of five diameters related with levels of severity of health effects. Location within the occupational environment is also known. Multivariate charts based on Hotellings T2 supplemented with graphics Shewhart and Bonferroni limits are applied as well as the generalization, to five variables, of the statistical decomposition of the Hotellings T2 for multiple comparisons. Some lines for further studies will be highlighted.