3G用户预测分析,是在数据仓库和数据挖掘技术迅速发展的基础上提出来的。对于3G的推出初期,电信运营商最想知道如何推广该业务,即哪些2G用户最有可能成为3G用户,最接近3G用户,从而提高命中率。
Abstract
Prediction and Analysis of 3G User, is proposed based on the rapid development of data warehouse and data mining technology. At the early introduction of 3G, telecom operators most want to know how to promote the business. In order to increase the percentage of hits, they want to know which kind of 2G users are most likely to become 3G users or be close to 3G users.
Based on a Hong Kong telecom operator’s data which includes more than 10,000 Samples, every sample has 250 variables include the data of Mobile use, demographic data and so on, we use CART decision tree software to make the existing 2G/3G users analyses and compared, classification tree generated. Then we extract the characteristics of 3G users.
Based on ranking the importance of the variables, we can get some of the important characteristics of 3G users. The model can be applied to find the 3G users in positioning from 2G users, or find 2G users who is most close to 3G users. Address this part of the population to promote, it will have better results than a random blind.
Key words: 3G user; Data Mining; Decision Tree; CART