STATISTICAL INFERENCE OF EXPONENTIAL RECORD DATA UNDER KULLBACK-LEIBLER DIVERGENCE MEASURE

Abstract Based on one parameter exponential record data, we conduct statistical inferences (maximum likelihood estimator and Bayesian estimator) for the suggested model parameter. Our second aim is to predict the future (unobserved) records and to construct their corresponding prediction intervals based on observed set of records. In the estimation and prediction processes, we consider the square error loss and the Kullback-Leibler loss functions. Numerical simulations were conducted to evaluate the Bayesian point predictor for the future records. Finally, data analyses involving the times (in minutes) to breakdown of an insulating fluid between electrodes at voltage 34 kv have been performed to show the performance of the methods thus developed on estimation and prediction.