Issue link: https://industrialengineer.epubxp.com/i/777540
February 2017 | ISE Magazine 53 the risk assessment of the ball-bearing system. CONTACT: Zhisheng Ye; yez@nus.edu.sg; 65- 66012303; Department of Industrial & Systems Engineering, National University of Singapore, 117576, Singapore Causation: A new opportunity to monitor multivariate categorical processes Categorical variables, also referred to as factors, prevail in today's manufactur- ing and service sectors. In many appli- cations, continuous measurements of quality characteristics are hard to obtain due to the intrinsic nature of variables or high cost for data collection. Moreover, it is often sufficient to use attribute levels to describe quality char- acteristics. When there are multiple categorical factors, the processes are known as multivariate categorical pro- cesses (MCPs). However, owing to the challenges and limitations in describing the correlation relationships among cat- egorical variables, there are few methods for monitoring and diagnosing MCPs. In many applications, causal relation- ships exist among categorical variables. Shifts upstream propagate downstream to affect variables based on the causal structure. For example, in a hot-form- ing process, the temperature is the cause of the material flow stress and the ten- sion in the workpiece. The latter two are also the cause of the fi nal dimension of the workpiece. If such cause-effect relationships among variables cannot be fully charac- terized and exploited, misleading con- clusions may be drawn. In such cases, a causation-based rather than correlation- based description would better account for the relationship among multiple cat- egorical variables. This provides a new opportunity for establishing improved monitoring and diagnostic schemes for MCPs. From this perspective, the monitoring of MCPs is investigated in "Causation- based Process Monitoring and Diagnosis for Multivariate Categorical Processes" by Jian Li from Xi'an Jiaotong Univer- sity, along with Kaibo Liu and doctoral student Xiaochen Xian from the Uni- versity of Wisconsin-Madison. They employed a Bayesian network approach to characterize such causal relationships and integrate the approach with statisti- cal process control methodology. Two control charts for detecting shifts in the conditional probabilities of multiple categorical variables that are embedded in a Bayesian network are proposed. The first chart provides a general tool, whereas the second chart integrates directional information that leads to a diagnostic prescription of shift locations. Both simulation and real case studies demonstrate the advantages of the causation-based monitoring and di- agnostic approaches over conventional correlation-based schemes. CONTACT: Jian Li; jianli@xjtu.edu.cn; 86-186290 13692; School of Management, Xi'an Jiaotong Uni- versity, Xi'an, Shaanxi 710049, China More research needed to find proper guidelines for sit-stand workstations Extended periods of seated work are an occupational reality for most office employees. Prolonged seated postures, This month we highlight two articles in Volume 4, No. 4 of IIE Transactions on Occupational Ergonomics and Human Factors (now IISE Transactions on Occupational Ergonomics and Human Factors). Examining the effects of computer configuration and sit/stand phase, the first article's researchers found that improper sit-to-stand workstation use can generate trade-offs in efforts to reduce musculoskeletal discomfort; more research is necessary to define "proper" sit- to-stand workstation guidelines for varying users. The second article investigates the effects of obesity on the balance response to a laboratory-induced trip. The research found not only a higher rate of falling among obese subjects but a deficient balance recovery response among the obese subjects who fell. Jian Li co-authored "Causation-based Process Monitoring and Diagnosis for Multivariate Categorical Processes." Kaibo Liu (right) and doctoral student Xiaochen Xian helped investigate the monitoring of multivariate categorical processes.