Using Neuro Fuzzy Methods for Creating Sales Forecasting Web Service

Abstract

This paper presents a new way of implementing the forecasting process using a neural network and a fuzzy controller. At the first stage of the process, the neural network performs the forecasting based on internal parameters, and then the fuzzy controller adapts the results to the changes of the external parameters. The main problem of neural networks in forecasting is the impossibility to foresee the changes in the external parameters. This problem is solved by the fuzzy controller, which reacts to any change in the external parameters and adjusts the forecasted data provided the neural network.