Invited Speaker---Prof. Su-Fen Yang


Department of Statistics, National Chengchi University, Taiwan, China

Biography: Prof. Su-Fen Yang is a distinguished professor of Statistics department at the National Chengchi University in Taiwan. She received the PhD degree in Statistics from the University of California, Riverside, USA. Her research interests are in the fields of statistical process control and quality management, Taguchi method, statistical computing, and applied Statistics. She has published over 170 application-oriented articles in academic journals and international conferences. Su-Fen Yang has received many research awards; research grants from government, industry, or school. She is a member of Quality Management Committee of Ministry of Economic Affairs Bureau of Standards, Taiwan; member of Committee of Chinese Society for Quality; head of research team in Commercial College and NCCU. She has served as the Department Head; Associate Editor of Journal of Quality; Associate Editor of Journal of Industrial Engineering Institute Association; Referee of many SCI, EI and international journals; Keynote Speaker, Invited Speaker, Co-Chair, Local Committee, Section Organizer, and Section Chair of international conferences.

Speech Title: Using A Simple Approach to Classify In-Control and Out-of-Control Process Variability
Abstract: From GE company's six sigma viewpoint, data variability is more important than data mean. To produce high quality products, variability of a critical quality characteristic should be maintained and monitored in the production process. We propose a new variance control chart based on a simple statistic to classify the in-control and out-of-control process variability for process data with non-normal or unknown distributions. We explore the sampling properties of the new monitoring statistic and calculate the average run lengths of the proposed variance chart. The average run length is biased for the monitoring statistic with an asymmetric distribution. That is, the ARL-biased control chart leads to take longer to detect the out-of-control process parameter than to trigger a false alarm. We herein propose an ARL-unbiased exponentially weighted moving average proportion chart to classify the in-control and out-of-control process variability. We also compare the out-of-control variance detection performances of our proposed variance chart with those of some existing variance charts.