Neural net based interregional telephone traffic models

This paper presents a neural network approach to modelling interregional telephone traffic. Real world observations in Austria have been used as a benchmark in our experiments. Single hidden layer feedforward connectionist networks have been designed to model interregional (domestic) telephone traffic flows between 32 Austrian regions. Remarkable success has been achieved in training the networks to learn the telephone traffic flows between the regions and in making accurate predictions. Our results show that the neural network approach is a leading contender to the conventional statistical approach.