The use of resampling techniques for lifetime data analysis in industrial engineering

This study concerns with estimating the parameters in lifetime of fragile population and the ratio of fragile population to the fragile and durable (mixed) population by using trunsored models (unification of truncated and censored models) approach. The purpose of this study is to illustrate the bootstrap resampling method used for the parameter estimation in trunsored models. The bootstrap method is especially convenient to make statistical inference when distributional assumptions are not valid. Therefore, trunsored models with bootstrapping, which follow a consistent strategy in statistical inference and data analysis, lead to more accuracy for evaluation. Like many real world cases, the thermal endurance data in material failure analysis do not follow any distribution perfectly. Furthermore, time and cost limitations prevent to observe a great number of data to analyze accurately. Thus, the trunsored model approach with bootstrapping is thought as potential to reduce the cost of destructive testing due to reduced frequency of testing, to prevent failures and to improve product reliability. The approach presented in this study may also be applied to many other real life problems.