Abstract Warranty data analysis is a form of life data analysis and is part of the engineering and business process of assessing product reliability and predicting the number of parts failing in the field and the associated warranty cost. However, oftentimes the prediction is started early and is based only on a few months of field data, which raises a question of data maturity. As data matures and the product is operating longer in the field the prediction changes, since the failure distribution becomes the function of the observation time. Data maturation has been known as a complicating factor in warranty data analysis affecting the accuracy of prediction, however there were very few attempts to look at how the data maturity affects the failure patterns and the probability of failure as the time in the field and consequently the observation time increases. This paper discusses the causes of data maturity, presents an analytical model to assess the maturation trends and presents several case studies based on the automotive electronics warranty data. The paper analyzes the patterns of how warranty data matures as more field data becomes available and how it affects the accuracy of prediction. It also suggests the criteria of determining the levels of data maturity.
展开▼
