Behavior-Based Telecom Tariff Service Design with Neural Network Approach

This paper presents a behavior-based telecommunication tariff service design model. Unlike traditional service design models, which includes demographic attributes to design telecommunication tariff services, this model only uses history data about telecommunication customer behavior, such as voice service usage, text service usage, and multimedia service usage to design new tariff services. An unsupervised algorithm, the self-Organizing Map neural network algorithm, is adopted to train the model. The model segments telecommunication customers into different groups only according to their behavior, service usage. The characteristics of customer service usage can be identified with the results. With the generated behavior-based groups and demographic attributes of customers as training data set, decision tree algorithm is adopted to reveal the characteristics of potential customers for the new designed tariff services.