Bayes Prediction in Exponential Life-Testing When Sample Size is a Random Variable

This paper deals with the problem of predicting the range in a future sample, based on ranges in the earlier samples. Samples are drawn from an exponential population. Sample sizes are taken as random variables and the prediction process is performed on the lines of Lingappalah (1978). First, for two samples, prediction intervals for two cases (sample sizes fixed, and random variables) are compared. Similarly, variances of predictive distributions for the two cases are compared. As an extension, prediction in k-th sample is made, based on earlier (k-1) samples. Finally, the case when all the parameters are equal is compared to the case when at least one is not the same.