Using a Hybrid Neural/Expert System for Data Base Mining in Market Survey Data

Victor Ciesielski, Gregory Palstra

This paper describes the application of a hybrid neural/expert system network to the task of finding significant events in a market research data base. Neural networks trained by backward error propagation are used to classify trends in the time series data. A rule system then uses these classifications, knowledge of market research analysis techniques and external events which influence the time series, to infer the significance of the data. The system achieved 86% recall and 100% precision on a test set of 6 months of survey data. This was significantly better than could be achieved by a system using linear regression together with a rule system. Both systems were able to perform analysis of the test data in under 5 minutes. The manual analysis of the same data took a human expert over four working days.

This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.