http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Jun Ding,Daqi Gao,Xiaohong Chen 보안공학연구지원센터 2015 International Journal of Smart Home Vol.9 No.3
Churn is a defection from a service provider. This study explores the role of social network in customers’ churn behavior .As the saying goes; customer is the god in service-based industries. Since the cost of customer acquisition is much greater than the cost of customer retention, customer churn prediction is more and more important. So far, research on churn prediction has involved user profile factors and social factors, but the analysis lacks interpretability and predictive horizon. In this paper, for a better understanding of churn behavior in MMORPG, we use longitudinal statistical models to examine whether the churn behavior in one person is associated with the churn of his friends on a complete data set of an entire society. Discernible clusters of active persons and inactive persons are visible in the network, and the relationship between people’s churn behavior extends up to three separations. Besides, we provide quantitative evidence for the strong ties hypothesis. The higher the number of common friends two persons have, the influence between them is more significant. At last, we use mixed effects cox models to predict churn in the future. Our study can provide a long forecast horizon, it is an obvious advantage to prevent the churn of a person.
Influence of Data Preprocessing
Changming Zhu,Daqi Gao 한국정보과학회 2016 Journal of Computing Science and Engineering Vol.10 No.2
In this paper, we research the influence of data preprocessing. We conclude that using different preprocessing methods leads to different classification performances. Moreover, not all data preprocessing methods are necessary, and a criterion is given to make sure which data preprocessing is necessary and which one is effective. Experiments on some real-world data sets validate that different data preprocessing methods result in different effects. Furthermore, experiments about some algorithms with different preprocessing methods also confirm that preprocessing has a great influence on the performance of a classifier.
Influence of Data Preprocessing
Zhu, Changming,Gao, Daqi Korean Institute of Information Scientists and Eng 2016 Journal of Computing Science and Engineering Vol.10 No.2
In this paper, we research the influence of data preprocessing. We conclude that using different preprocessing methods leads to different classification performances. Moreover, not all data preprocessing methods are necessary, and a criterion is given to make sure which data preprocessing is necessary and which one is effective. Experiments on some real-world data sets validate that different data preprocessing methods result in different effects. Furthermore, experiments about some algorithms with different preprocessing methods also confirm that preprocessing has a great influence on the performance of a classifier.