Automatic Identification of Time Series Features for Rule-Based Forecasting

13 Pages Posted: 7 Feb 2005 Last revised: 1 Jan 2012

See all articles by Monica Adya

Monica Adya

Marquette University - College of Business Administration

Fred Collopy

Case Western Reserve University - Department of Information Systems

J. Scott Armstrong

University of Pennsylvania - Marketing Department

Miles Kennedy

Independent Author

Abstract

Rule-based forecasting (RBF) is an expert system that uses features of time series to select and weight extrapolation techniques. Thus, it is dependent upon the identification of features of the time series. Judgmental coding of these features is expensive and the reliability of the ratings is modest. We developed and automated heuristics to detect six features that had previously been judgmentally identified in RBF: outliers, level shifts, change in basic trend, unstable recent trend, unusual last observation, and functional form. These heuristics rely on simple statistics such as first differences and regression estimates. In general, there was agreement between automated and judgmental codings for all features other than functional form. Heuristic coding was more sensitive than judgment and consequently, identified more series with a certain feature than judgmental coding. We compared forecast accuracy using automated codings with that using judgmental codings across 122 series. Forecasts were produced for six horizons, resulting in a total of 732 forecasts. Accuracy for 30% of the 122 annual time series was similar to that reported for RBF. For the remaining series, there were as many that did better with automated feature detection as there were that did worse. In other words, the use of automated feature detection heuristics educed the costs of using RBF without negatively affecting forecast accuracy.

Keywords: Rule-based forecasting, time series, forecasting, heuristic forecasting

Suggested Citation

Adya, Monica and Collopy, Fred and Armstrong, J. Scott and Kennedy, Miles, Automatic Identification of Time Series Features for Rule-Based Forecasting. International Journal of Forecasting, Vol. 17, 143-157, 2001. Available at SSRN: https://ssrn.com/abstract=662581

Monica Adya

Marquette University - College of Business Administration ( email )

P.O. Box 1881
Milwaukee, WI 53201-1881
United States

Fred Collopy

Case Western Reserve University - Department of Information Systems ( email )

10900 Euclid Ave.
Cleveland, OH 44106-7235
United States

J. Scott Armstrong (Contact Author)

University of Pennsylvania - Marketing Department ( email )

700 Jon M. Huntsman Hall
3730 Walnut Street
Philadelphia, PA 19104-6340
United States
215-898-5087 (Phone)
215-898-2534 (Fax)

HOME PAGE: http://marketing.wharton.upenn.edu/people/faculty/armstrong.cfm

Miles Kennedy

Independent Author

Register to save articles to
your library

Register

Paper statistics

Downloads
76
Abstract Views
813
rank
314,689
PlumX Metrics