|
Abstract:
|
It has been previously shown a purely sequential method for approximating the optimal sample size is better than other sequential methods sequential, for example, the exponential case. In this paper, we develop a purely sequential risk estimation procedure for estimating the mean of a gamma process. Because, in this case, the sequential procedure uses the combined sample from all stages, the estimator and the sequential sample size are dependent. We show that our method guarantees that we have an exact bounded risk estimate. These methods are demonstrated through extensive simulation studies and application to real data sets.
|
