A Study of Service Center Records Using Data Mining

Abstract

In many manufacturing companies, large databases containing data on failures and repairs of products are maintained. This paper reports on a case where information from such a database is extracted. Such information extracted could be useful for identifying opportunities for reliability improvements in the products. The database was obtained from a manufacturer of inkjet printers and contains both fixed-format and free-form text fields. At present, techniques developed in data mining and information retrieval are mainly designed to handle either fixed-format or free-form text fields, but not a combination of both. The approach taken in this paper is to first transform the free-form text fields into a number of fixed-format fields through analysis of the frequency of key words. Association analysis was then carried out on the resulting fixed-format fields. Results obtained from the analysis produced a number of associations that could contribute to improving product reliability.