Predictive Density Estimation in Accelerated Life Testing for Lognormal Life Distributions

• Activity Number: 133
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Quality and Productivity Section
• Author(s): Ananda Jayawardhana*+ and V. A. Samaranayake
• Companies: Pittsburg State University and Missouri University of Science & Technology
• Keywords: Profile Predictive Likelihood ; Maximum Likelihood Predictive Density ; Prediction Bounds

Abstract

This paper proposes a method of obtaining a predictive density for a future observation from a lognormal distribution, using data from an accelerated life test. It is based on the Maximum Likelihood Predictive Density method proposed by Lejeune and Faulkenberry (1982). The resulting predictive density is a t-distribution. A Monte Carlo simulation study was carried out using parameter values motivated by the data published by Crawford (1970). Monte Carlo methods show that the predictive density provides liberal prediction bounds but a simple modification to the degrees of freedom yields slightly conservative coverage probabilities. Simulation results and an example will be presented.