Warranty prediction during product development: Developing an event generation engine in an engineer-to-order environment

In order for manufacturing companies to stay competitive, it is necessary to drive warranty system improvements in terms of improved product reliability, improved service delivery efficiency and properly designed warranty policies. However, traditional methods for assessing warranty performance are not always sufficient to alert product development teams of the impending warranty issues. Furthermore, improved assessment methods are needed to aid product development teams make decisions related to the warranty performance of the product. The focus of this research was to develop a framework to integrate statistical inference methods and data mining techniques to create a warranty event generation framework. This was done on the context of an engineer-to-order product development environment. The objectives of this work were: (1) to develop an inference model for the integration of disparate data sources; (2) to demonstrate that multiple data streams can be conditioned for input into the above inference model; (3) to develop the above model and process in light of actual data. This thesis will report on the progress and challenges that have been made toward fulfilling these objectives. The thesis closes by outlining the future research agenda for developing a warranty event generation engine that can integrate data from disparate data sources