Rule-based Forecasting (RBF) integrates statistics and domain knowledge to deliver more accurate forecasting techniques. RBF is an expert system that uses features of time series to select and weight extrapolation techniques. The current expert system consists of 99 rules that combine forecasts from four simple extrapolation methods – random walk, linear regression, Holt’s exponential smoothing, and Brown’s exponential smoothing. In this sense, RBF is a knowledge based system that successfully combines domain knowledge with statistical techniques.

Results from independent validations and the recent participation in M-3 Competition have found that RBF is consistently more accurate than leading benchmarks such as random walk and equal weights combining.

This following set of slides provides a summary of RBF and presents results from some validations:
- An Overview of RBF (PowerPoint)
- An Overview of RBF (PDF)


This website is the product of collaboration on the part of J. Scott Armstrong and This email address is being protected from spambots. You need JavaScript enabled to view it. who is responsible for the maintenance of this page.

Monica Adya
Assistant Professor of Management
College of Business Administration
Marquette University
Milwaukee, WI 53201-1881
Telephone: (414) 288-7526