Kalman and Moving Average Filters for Forecasting: Systematics of Demand Processes and Extensions.
Abstract
This technical report is a digest of theoretical (mostly) results on short term forecasting. The ubiquitous criterion is mean square error MSE, which is the most amenable to mathematical analysis, but in some transformations of process variables to be forecasted infer other error measures. This study is also a proselytization for the Kalman filter, which together with its underlying vector process model, is a powerful and flexible algorithm which encompasses many other techniques. To an extent, the study concerns systematics, a scheme for classifying such well known forecasting algorithms as exponential smoothing, moving averages, regression line fitting, by the underlying processes for which these algorithms are optimal or sub-optimal.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1976
- Accession Number
- ADA032496
Entities
People
- Donald A. Orr