Takenori Takahashi, Visiting Professor, Graduate School of Keio University
There are two types of education that handles data: science (basic science) and engineering (applied science). This study focuses on the latter, and discusses the education of problem solving and task achievement, which are important themes. There are significant differences between the two (detailed in the presentation), but what they have in common is that the first half is analysis (modeling) and the second half is design (optimization), and it is important to adopt simulation experience education.
Analysis (modeling) can be divided into cases where the target is transparent in advance (when the factors, levels, and structure are known) and cases where the target is opaque (when the factors, levels, and structure are only vaguely known). Custom designs are effective for the former. On the other hand, identification, which involves conducting initial experiments to determine the factors, levels, and structure, is essential for the latter, and this study proposes a method of identification. After identification, additional experiments are required for modeling, and in this case, extension designs are effective.
In design (optimization), mathematical programming (solving by constraints and objective functions) is used based on the created model. Using the profile function is effective in this case.
This study introduces the above theory and a simulation experience education using a flying ball simulator, a virtual experiment that runs on JMP, as well as other simulators in order to learn this theory in a practical way.
