Analysis of MUF Data Using ARIMA Model

Th& analysis of MUF has taken several steps from the pioneering work of Jaech[l] who introduced several statistical methods to evaluate MUF data, to Stewart[2] with his minimum variance unbiased estimator techniques, and most recently the application of Kalman Filtering to detect losses in MUF data by Pike and Morrison[3]. The references above as well as others have presented techniques to detect losses using inventory and transfer data. In this paper we present a new technique for estimating the loss when the loss scenario is known. This technique differs from the others in that it applies the Box- Jenkins[4] time series analysis to model the stochastic process (the observed MUFs) and uses the one-step-ahead forecasts to indicate whether a loss is occurring or not. The purpose of this paper is twofold: 1) to introduce the Box-Jenkins time series analysis methods anc 2) to use these techniques in determining whether a significant loss has occurred.