Warranty Policy Determination using Data-Mining Association Rules (Case study: Electronic Facilities Company)

Abstract: Nowadays, varieties of warranties are presented by manufacturers. Improving the warranty policy imposes some costs on the producers. As a result, one needs to rely on factual and reliable data as well as the data on defects and repair when it comes to making warranty policy. To this end, this study uses data mining method. That is, an electronic equipment company's warranty data including 3500 defects reports within a 5-year period sample were mined, using association rules. This yielded significant patterns based on the data. Out of total derived rules and patterns, some rules describe the associations between the products and their defects and repairs better than others. So, having these information, we could determine the warranty policy for 24 products. The findings of this study can reduce warranty costs via optimization of warranty policies decisions. Introduction: Buyers of products want assurance that the product will perform satisfactorily over its useful life when operated properly. This is achieved through post-sale support (also called product support) provided by the manufacturer. Warranty is one of post-sale supports that serves as a way to promote the competiveness capacity of the products. The complex competitive market and customers' demands have increased the competitions among the manufacturers in order to provide more customers with better warranties. Consequently, nowadays, varieties of warranties are presented by manufacturers. Effective management of product warranty requires proper evaluation of alternative warranty policies (Blischke & Murthy, 1992). Offering better warranty policies conveys greater assurance to buyers and can result in greater sales. However, this increases the cost of servicing the warranty. As a result, one needs to rely on factual and reliable data as well as the data on defects and repair rather than estimation and guess when it comes to making warranty policy. A producer can use the data gathered during the warranty period (generally called “warranty data”) for various purposes. Since warranty data features a variety of failure modes, it can activate an early warning for design errors, highlight faults in the manufacturing process, and help enhance a product by understanding customer usage profile. This information can also help in estimating future expenses (Jeon & Sohn, 2015). Warranty data are strictly confidential for most companies because they relate to product quality, reliability, and are therefore critical to consumers’ product goodwill (Buddhakulsomsiri & Zakarian, 2009). Materials and Methods:Product quality problems are monitored during the warranty period through the claims filed against the products. This process generates large volumes of warranty data records, such as product problems in the form of repair related labor codes, problem descriptions, actions taken, repair dates, and repair costs (labor and parts). Analyses of these data records may provide significant benefits to product manufacturers (Buddhakulsomsiri & Zakarian, 2009). To this end, this study uses association rule method. The association rule (AR), which is a data-mining method, is used to determine the degree of relevance between variables (Jeon & Sohn, 2015). In this paper, an electronic equipment company's warranty database including 3500 defects reports and resulting warranties within a 5-year period sample were mined, using association rules. This yielded significant patterns and rules based on the data. Data processing is done using SPSS Modeler 14.2 software. Results and Discussion: The results were obtained from the implementation of the model by the software, including 475 association rules. Out of total derived rules, 72 rules which describe the associations between the products and their defects and repairs better than others, were selected. These rules clarify the relationship between various products and their types of defects, the intensity of the defect, the number of the defect, the repeatability of the defect, reparability and the repair costs. This information provides the knowledge needed to decide on all variables in a warranty policy. And having this information, we could determine the warranty policy for 24 different products, in 4 categories of ‘warranty period’, ‘warranty cost’, ‘Compensation method’ and ‘warranty dimensions’. Conclusion: The findings of this study can decrease warranty costs via optimization of warranty policies decisions. Because, implementing these warranty policies reduces the manufacturer's risk of warranted products and reduces the cost of warranty service. Consequently, companies not only use the huge amount of stored data that contains valuable information about the various product failure and product warranties, but also will be interested both in the promotional benefits of the warranties in attracting customers, and in the benefits of reduction of the warranty costs by providing a good warranty policy for their products. This finally leads to increased profitability of the organization while achieving competitive advantage. The producers are recommended to reduce the warranty costs and to increase the profitability of production industries, using the proposed method as well as numerous data to make warranty policies in different firms. References Blischke, W.R. & Murthy, D.N.P. (1992a) “Product warranty management – I: A taxonomy for warranty policies”, European Journal of Operational Research, 62, 127–148. Buddhakulsomsiri, J. & Zakarian, A. (2009) “Sequential pattern mining algorithm for automotive warranty data”, Computers and Industrial Engineering, 57)1), 137–147. Jeon, J. & Sohn, S.Y. (2015) “Product failure pattern analysis from warranty data using association rule and Weibull regression analysis: A case study”, Reliability Engineering and System Safety, 133, 176–183