Neural networks vs. regressive models – forecasting of demand on three kinds of goods

Abstract

Neural networks are a relatively young developmental area. They represent an important tool for solving various problems from different fields. For this reason they are interesting for solving management problems. This paper gives an answer to the following basic question: Is the (feedforward) neural network a better method for forecasting (food retail) than the traditional regressive method? The results have shown that the regressive method became a competitive method only when the regressive model reached a high explanatory variance. When the regressive model had a lower explanatory variance the neural network proved to be a much better method than the regressive method. The method of neural network has proved to be a better method for forecasting the food retail then the traditional regressive method. This has additionally been confirmed by the results of estimating methods and conclusions of same researches.