Configuration of Daily Demand Predicting System Based on Neural Networks

Having a clear vision about future demand is a crucial key to enhance the commercial competitiveness to any efficient supply chain. However, demand forecasting is certainly not an easy task for a manager who had the choice between using traditional forecasting techniques encompassing time series methods, causal methods or simulation methods, or techniques based on artificial intelligence like artificial neural networks (ANNs), fuzzy logic or adaptive neuro fuzzy inference system (ANFIS). This paper focuses on the implementation and configuration of the artificial intelligence of neural networks, and more precisely the multi layer perceptron's structure, as a prediction system to produce daily demand forecasts based on historical demand information. The results indicate that adding new inputs to the neural network, in our case study, has a positive impact on the accuracy of the short term demand forecasting. In the numerical experimentation, the effectiveness of the proposed model is validated using is validated using a real-world data of a leader supermarket in Morocco.