Research On Classified Mobile Users’ Service Choices Through Fuzzy C-means- Multinomial Logit Model

With the development of the telecom industry and the competition between the telecom operators, telecom operators continue to introduce complex products and service categories to attract the consumer.It is more difficult for consumers to find suitable products within the complex types and similar property of telecom products.During make the product selection, telecom users have problem to consider their own needs reasonably and choose the products that meet their needs from a variety of products, or they can select the reasonable choice after the consumer experience in the period of time. On the other way, telecom Company can not have a precise prediction for the need of users. To solve this problem, research for qualitative analysis of users’ need and choice of telecom behavior analysis have been developed. But since the study data is bulky and the internal properties are complex, the forecast for product selection of the telecom users’ demand is not accurate and cannot be measured from the quantitative point of view. Therefore, study the reasonable service selection of the telecom users and seek the scientific method of service selection, not only can promote the users consumption reasonably, but also promote the telecom company to grasp the market demand accurately and make more reasonable product strategy.The paper summarizes the domestic and foreign research of telecom users’ product behavioral. Combine the theory of consumer choice with the telecom users’ features, the paper explains clearly that construct the multinomial logit model after the customer segments can effectively improve the prediction accuracy.By compare different algorithms and discrete choice models, the paper selects fuzzy C-means and multinomial logit model as the base model, and the combination of the model is applied to the mobile users’ service selection among research. In order to study the service selection under the premise of limited rationality, the paper compares the price of the package and the actual cost and makes it as a standard to measure whether the users consume reasonably. Through theoretical analysis, the paper explains clearly that construct the fuzzy C-meansmultinomial logit model can improve the prediction accuracy and the results show that the model constructed on the reasonable users can predict the users’ demand more reasonably and the prediction accuracy of the model constructed on the classification samples is higher than the overall sample selection model.