Estimating Multivariate Conditional Distributions — An Application to the Truck Sales Forecast

A concept for forecasting the conditional multivariate distribution has been developed. It allows the forecast of the joint distribution of target variables in dependence on explaining variables. The concept can be applied to general distribution families such as stable or hyperbolic distributions. The conditional distribution parameters are estimated by a global optimization method, using neural networks for functional approximation. The information about a complete distribution of forecasts can be used to quantify the reliability of the forecast. A comparison with conventional forecasting concepts is done and the additional benefit of forecasting conditional distribution in general, and of hyperbolic distribution in particular is shown. The concept is illustrated on a case study concerning the future truck demand. In this application, th e distribution parameters are conditional on properties of the product and information about existing orders.