Modelling Future Demand by Estimating the Multivariate Conditional Distribution via the Maximum Likelihood Principle and Neural Networks

E.A. Stützle and T. Hrycej (Germany)


Forecast of Probability Distribution, Uncertainty Management, Probabilistic Reasoning, Neural Networks, SpareParts Demand Forecast


A new concept for modelling and forecasting is introduced. The maximum likelihood principle is used to identify the underlying multivariate conditional distribution. The dis tribution parameters are conditional on input features such as properties of the product. The conditional distribu tion parameters are estimated by a global optimization method, using neural networks for functional approxima tion. The goal is to construct a general attribute-based fore cast model, which can be applied to novel cases with new attribute combinations. The information about a complete distribution of forecasts can be used to quantify the relia bility of the forecast. The reliability information is partic ularly useful for decision support, e.g. if the forecast error causes strongly asymmetric costs. This is illustrated on a case study concerning the spare parts demand forecast.

Important Links:

Go Back