Sample Size Determination and Confidence Interval Derivation for Exponential Distribution

摘要

Exponential distribution is the simplest and the most common useddistribution for risk and reliability data analysis. This paper investigatesthe distribution of the estimator from maximum likelihood estimation(MLE). This estimator distribution shows the MLE estimator biasedand it should be corrected for small data size problem. The deriveddistribution is applied to determine the optimal sample size forexperiment design, and it is applied to derive confidence interval. Thisconfidence interval is compared with confidence interval derived fromFisher information matrix and the likelihood ration method.