The Pyke estimator and the exponential distribution
Hugh Allan Kelley · Calhoun: The Naval Postgraduate School Institutional Archive (Naval Postgraduate School) · 1973
It is common to estimate a distribution by means of a step function. Such estimates can be made continuous by connecting the left points of the steps with straight line segments. In this paper, the best estimator of this class is found for data which is exponentially distributed using minimum risk. The risk is then compared with those of the sample distribution function and the Pyke estimator.