Application of model intelligent selection to tobacco market forecast

Accurate forecast is the key for sales decision-making of tobacco commercial companies. Model selection plays an important role in sales forecast. This paper adopts expert system theory to model selection. Model knowledge tree is set up by selection rules to make knowledge structured and systematic. On the basic of model selection knowledge tree, build up model selection reasoning machine adopting framework reasoning and local-preference searching strategy. The method is applied successfully in marketing decision support system for a tobacco commercial company.