A fuzzy neural network model with application to telecommunications business

There are many factors involved in a revenue-forecasting problem of telecommunications business (TB), which are not independent and conflicted with each other. The traditional methods of forecasting TB revenues are based on regression analysis, but the precisions of the results are usually not satisfied in practice. In this study, statistical analysis is first used to evaluate factors and then the main factors are selected as inputs to construct a neural network model combined with fuzzy rules. Furthermore, the fuzzy neural network (FNN) model is applied to predict the revenue of a specific telecommunications business. The results indicate that the developed FNN model satisfies practical requirements and is possible to expand to other business applications, such as customers' evaluation and financial risk pre-warning. To make this clearer, an illustrative example is given for demonstration.