Canadian Forest Service Publications
Diagnosis of sparse adoption data using an expert system-guided innovation diffusion simulation model. 2008. Thomson, A.J. The Innovation Journal 13(3): article 11.
Issued by: Pacific Forestry Centre
Catalog ID: 29212
CFS Availability: PDF (download)
Forecasting of the likely rate of uptake of an innovation, idea or product is at the heart of business case development, but adoption of simulation models to produce such forecasts is often low due to complexity faced by users in selecting appropriate sets of parameters that provide a best fit to available observations. In addition, when observations are sparse, correlation between observed and predicted values is not an appropriate guide to model appropriateness. The present study describes (a) a computer simulation of the innovation diffusion process, and (b) an analysis and diagnosis package that can be added to simulation models to interpret output and advise on parameter changes to bring model scenarios in line with observations. The approach is twopart. First, envelopes are created around different scenario outputs and the relationship of the observation set to the envelopes determined. Subsequently, a rule mapping table edited by an expert user, links parameter modification options to specific aspects of the observations-envelope relationship. The expert system thus acts as a change agent to reduce the complexity of simulation model use.