Forecasting The Telecommunications Services Volume Based On Artificial Neural Network

This paper established the telecommunications services volume forecasting model and forecasted the development of the telecommunications industry based on the analysis of the major impact of the factors.The paper expounded development of the telecommunications industry in China firstly, and pointed out the characters of telecommunications services volume's data, such as non-linear, small- sampled and complexities of factors affecting to each other.The paper introduced some commonly forecasting methods. Such as the return of specific prediction method, smoothing index, grey model prediction, BP neural network, RBF neural network .Each method based on different mathematical theory and model, so that they have different data requirements, different predicted accuracy and different predicted periods.The traditional methods in the telecommunications industry forecast based on the analysis of factors. Based on this, we always set out the predicted return equation. But in the following circumstances such as it had little data, the affecting factors' relationship was complex and opaque, the result was not very good. So this paper used China's telecommunications services volume of historical data between1989-2005 for example, based on the developing characteristics and the main factors, predicted the services volume on the different methods. In order to reflect the comparability of the forecasting methods, we only tested it on 2005 Telecommunications volume. And we can compare the predict accuracy of different methods.By comparing various forecasting methods, this paper summed up the advantages and disadvantages of various forecasting methods, analyzed the various predicting methods for the predicting periods and the accuracy of the results .The final result showed that the prediction of the telecommunications services volume which was based on regression analysis of the neural network having the best effect. At last, the paper used the model to predict the next five years' telecommunications services volume and made some proposals to the development of the industry.