基于Boosting的市场值函数算法及其评价
Keywords: Boosting算法 ,市场值函数 ,lift值
Abstract:
发现具有潜在市场价值的客户群是直销中的一个关键问题，尽管一些标准的数据挖掘算法可以用来解决此问题，但效果并不理想．为此，采用市场值函数算法，它以信息论为基础，通过构造一个线性市场值函数来对客户进行排序，从而发现最具有潜在市场价值的客户群．实验结果表明，它的评价值可达80％以上，并且具有很好的可解释性．同时，将Boosting算法应用到市场值函数算法中，用以提高市场值函数的预测效果；在3个不同的数据集上进行了实验，评价值均提高了一个百分点．
References
[1] TOM M M. Machine Learning[M]. New York: The McGraw-Hill Company, Inc, 1997. [2] YAO Y Y, ZHONG N. Mining market value function for targeted marketing[A]. 25th Annual International Computer Software and Applications Conference (COMPSAC' 01) [C]. Chicago: IEEE Computer Society Press,2001. 517-522. [3] 黄佳进，刘椿年，李文斌．市场值函数挖掘的研究和实现[J]．北京工业大学学报，2003，29(1)：94-97． [4] HUANG Jia-jin, LIU Chun-nian, LI Wen-bin. Research and implemenitation of mining marker value function[J]. [5] Journal of Beijing University of Technology, 2003, 29(1): 94-97. (in Chinese) [6] ROBERT E S, YORAM S. Improved boosting algorithms using confidence-related predictions[J]. Machine Learning,1999, 37(3): 297-336. [7] ELKAN C. Boosting and Naive Bayesian Learning[R]. San Diego: University of California, 1997. [8] DRUCKER H, CORTES C. Boosting decision trees[J]. Advances in Neural Information Processing Systems, 1996,8: 479-485. [9] LING C X, Li C H. Data mining for direct marketing: Problems and solutions[A]. Proceeding 4th International [10] Conference on Knowledge Discovery and Data Mining[C]. New York: AAAI press, 1996. 73-79. [11] KAYMAK U, SETNES M. Target Selection Based on Fuzzy Clustering: A Volume Prototype Approach to Coil [12] Challenge 2000[R]. Leiden: Amsterdam and Leiden Institute of Advanced Computer Science, 2000. 1-6.
Full-Text
