A decision support system for sales forecasting through fuzzy neural networks with asymmetric fuzzy weights

摘要

Sales forecasting plays a very prominent role in businessstrategy. Numerous investigations addressing this problem havegenerally employed statistical methods, such as regression orautoregressive and moving average (ARMA). However, sales forecastingis very complicated owing to influence by internal and externalenvironments. Recently, artificial neural networks (ANNs) have alsobeen applied in sales forecasting since their promising performancesin the areas of control and Pattern recognition. However, furtherimprovement is still necessary since unique circumstances, e-g-,promotion, cause a sudden change in the sales pattern. Thus, thisstudy utilizes fuzzy logic a proposed fuzzy neural network (FNN) forthe sake of learning fuzzy IF--THEN rules obtained from the marketingexperts with respect to promotion. The result from FNN is furtherintegrated with the forecast from ANN using the time series data andthe promotion length through the other ANN. Model evaluation resultsindicate that the proposed system can more accurately perform thanthe conventional statistical method and single ANN.