List of works
Conference presentation
Stochastic Process Modeling in a Variable Parameter Framework
Date presented 06/1983
Pragmatic Research: International Symposium on Forecasting, 06/05/1983–06/08/1983, Philadelphia, Pennsylvania, USA
A common assumption with time series modeling is that the parameters of the process remain unchanged as the series realization grows. This paper examines the case in which the parameters of a stochastic process the time series model are allowed to vary systematically within a given realization. Specifically, a model is developed that explicitly considers the structure of parametric drift within a given stochastic process. It is shown that models of this type are easily developed and add an element of enhanced precision to the predictive performance of time series forecasts.