An artificial neural network approach for improved demand estimation of a cool-disk manufacturer

Neural networks have successfully been used for demand forecasting. However, due to a large number of parameters to be estimated empirically, it is not a simple task to select the appropriate neural network architecture for a demand forecasting problem. So, in this paper, we examine the effects of the number of input and hidden nodes and hidden layers as well as the size of the training sample on the in-sample and out-of-sample performance. Also, we consider a new forecasting approach inspired from the regression method for weekly demand forecasting as a benchmark for comparison with ANN. It is shown that ANN is superior to a regression model through analysis of variance and well-known relative error estimation methods. This is the first study that presents an integrated ANN approach for improved demand estimation in a cool-disk manufacturer company.