Telecommunication user classification method based on iterative decision tree

Abstract

The invention relates to a telecommunication user classification method based on an iterative decision tree. The method comprises the steps of (S1) constructing a user social relationship network G(V,E) according to basic information, call and short message records of a user, (S2) extracting characteristics from the G(V, E) or updating characteristics of other users according to classified users,(S3) classifying unclassified users in G(V, E) by using an iterative decision tree, judging whether a condition that |sFriend-nFriend| is larger than or equal to L is satisfied or not, classifying ifso and recording a number of classified users as k, and going to the step (S2) if k is larger than K, otherwise, going to step (S4), (S4) allowing L=L-1, if L is larger than or equal to 0, returningto step (S2), otherwise, going to step (S5), and (S5) outputting a classification result. A very low error rate can be achieved in the case of few training sets, the influence of over-fitting caused by the high proportion of the training sets can be overcome, and the accurate positioning of the users is achieved.

Description

technical field [0001] The invention belongs to the fields of artificial intelligence and data mining, and in particular relates to a method for classifying telecommunication users based on an iterative decision tree. Background technique [0002] 1. Mobile Marketing Technology [0003] Mobile marketing has been widely studied in recent years. On the one hand, some people have theoretically studied why consumers can accept mobile phones as a way to promote information. Through empirical research, they found that when using mobile client for mobile marketing, the key to whether it can be accepted by users lies in the entertainment and value of the information provided to customers. On the other hand, some people have studied the behavior of mobile phone users in Italy in an empirical way, based on factors such as consumers / users' lifestyles, motivations for use, and product / service attributes, through element analysis and cluster analysis methods. Others surveyed consumers'...

AI Technical Summary

Problems solved by technology

However, at present, when telecom operators classify users, they are mainly based on the user's personal information, such as name, gender, location, etc., and do not make full use of the operator's own rich data such as user call and SMS records, which leads to the accuracy of traditional classification methods. lower sex

Images