Level: Intermediate
【Presenter】
Akira Hososhima, Representative of Tobayashi Consulting
[Announcement summary]
The text mining method has some application examples that should be noted even by undergraduates, such as analysis of quality trends and reports of text mining of maintenance records to optimize maintenance. In this report, we will introduce an example of finding the problems of quality and productivity by text mining the production record data and investigating the cause.
The production performance data includes several variables such as machine and operator numbers, addition conditions, material conditions, and special notes. Special notes are text data manually input by the operator when an irregular event occurs, and since the content contains words unrelated to the abnormal event, register it as a stop word and exclude it from the analysis target. Then replace synonyms such as "mistakes, mistakes, mistakes" with one word by recoding. Through such pretreatment, the presence or absence of frequent abnormal words is stored in a column, and the presence or absence is analyzed by the relationship between the production condition and the bivariate. The correlation with was clarified, and the cause of the abnormality could be investigated.
