DETECTING A TYPICAL RUNS USING THE BETA DISTRIBUTED   CHART IN PHASE I PROCESS DATA – THE TOBACCO CASE STUDY 

Chisimkwuo JOHN

Department of Statistics, Michael Okpara University of Agriculture, Umudike, Abia State, Nigeria.

Abstract

The Hotelling’s  is one of the most important multivariate procedures in the study of statistics. Its usage in the process industry becomes apparent as many modern industrial processes are highly multivariate in nature. In this study, a variant of the  chart notably the Beta distributed  is shown to have the ability to detect atypical runs during Phase I clean up processes. This study uses an exploratory 3-Dimensional scatterplot system to visualize and confirm the presence of atypical runs in a process data set, and then the Beta distributed   was employed to showcase these atypical runs. An empirical study using the tobacco data set showcases this scenario and at the end, the use of the Hoteling’s  is recommended in Phase I multivariate statistical process control operations as a viable data cleaning approach especially when removing atypical Phase 1 observations.

Keywords: Statistical Process Control (SPC), Hotelling , the Scatter plots, rgl