Data Analysis Based on Warranty Database

Abstract

While product quality and reliability are important to a manufacturer’s success, it is clearly desirable to minimize the amount of data that must be maintained in order to evaluate reliability. Too little data can compromise the manufacturer’s ability to analyze reliability while maintaining data unnecessarily is costly. For many industrial products, sales information comes from the database of the sales department and warranty records come from that of the maintenance department. That is, the data come from different sources. Thus, for specific products sold in a month, this database gives neither the exact number of failures at any month, nor the actual operating time (e.g. vehicle mileage or photocopy volumes) to failure. Furthermore, exact monthly sales figures for a particular product are not always available. Manufacturers encounter difficulties in obtaining detailed information on reliability from such limited databases. This paper investigates a “minimal” and “sufficient” database to estimate product reliability