Level: Intermediate
In today's world, where work-style reform is a hot topic, the importance of streamlining experiments and analyses is increasing. To demonstrate the power of JMP in streamlining experiments, it is good to show how the number of experiments can be drastically reduced by replacing experimental data with definitive screening designs (DSDs) or custom designs. The response data in this case should be extracted from existing experimental data. When experimental data is found to be divided into multiple tables, it should be combined into a single table, multivariate analysis should be performed, and the results visualized with a profile to make the audience aware of the pitfalls of OFAT (One Factor at a Time) methods. For the approach of analyzing repeated experimental data using the mean, it should be shown that there are methods such as cumulative processing, multi-objective optimization using mean and variance, and robust optimization.
In experimental design, it is often difficult to predict the existence of interactions, and interactions are not uncommon. DSD avoids confounding between main effects and interactions (2FIs) or between 2FIs, and requires only about twice the number of experiments as the number of factors. This is a significant advantage. This report will cover what we have learned from actually using DSD, the breakdowns that occur when the number of main effects plus interaction terms approaches the number of factors, solutions using augmented designs, and points that seem to be of practical importance from DSD-related papers obtained from the JMP community and ASQ.
After serving as General Manager of FA Development, Director and Head of Research and Development, Director and Head of Quality Assurance Promotion at Yamato Honeywell (now Azbil), and Advisor to Azbil Kimmon, he established Torin Consulting. His areas of expertise include yield and quality improvement through production data analysis, statistical problem solving in general such as market defect forecasting, robust design, optimized design, and experimental design, and on-site guidance such as design review, root cause analysis (RCA), prevention of human error, and process improvement. His publications include "The Essence of Internet Business" (JUSE Publishing, 2001, co-authored) [Telecom Social Science Award winner], "Practical Success Strategies for Venture Companies" (Chuo Keizai-sha, 2011, co-authored), and "An Easy-to-Understand Book on Problem Solving" (Nikkan Kogyo Shimbun, 2014, sole author). His main paper is "A Proposal for a Workshop to Promote the Dissemination and Acceptance of Awareness Regarding Near Misses and Anxiety on Production Lines" (Japan Society for Quality Control, 2016, Quality Technology Award winner). Main presentation: "Proposal for a mechanism to visualize organizational factors that induce work errors and promote improvement" (Discovery-Japan 2018)
"Solving Quality Problems with JMP: Failure Analysis and Reliability Prediction in Manufacturing" (Discovery-Japan 2019)
