A minimum cost exponential cusum test |
| |
Authors: | Amy Hancock Boyd |
| |
Institution: | Department of Mathematics and Computer Science , Colby College , Associate Professor, Waterville, 04901-8858, ME |
| |
Abstract: | A minimum cost CUSUM test for an event rate increase when inter-event times are exponentially distributed is presented. Optimal values of the test decision parameters, h and k, are developed from a renewal reward model of the event cycle by combining a non-linear optimization technique with an exact method for determining exponential average run lengths. Test robustness for event cycle parameter estimates and departures from the assumption of exponentially distributed inter-event times are discussed in the context of an injury monitoring scenario. Robustness to positively serially correlated observations emanating from EAR(1) and EMA(1) processes is also examined. |
| |
Keywords: | exponential inter-event times CUSUM test expected cost monitoring serial correlation |
|
|