基于游戏玩家流失预警的改进决策树算法  

Improved decision tree algorithm based on game players erosion warning

作　　者：林雪云[1] 

机构地区：[1]福建师范大学福清分校,福建福清350300

出　　处：《长春工业大学学报》2016年第2期182-186,共5页Journal of Changchun University of Technology

基　　金：福建省教育厅B类项目(JB13197)

摘　　要：首先对样本数据的影响因素进行logistic回归剔除绝大部分噪声因素,再通过降维和共线性诊断进一步筛选出影响因素中对实际结果有重要影响的指标以及共线性的指标,剔除无用和重复的类别,对清洗后的数据进行分析。最后进行了实例验证。First the logistic regression is used to eliminate the noises from the important factors in sample data,and then the influential and collinear indexes in the factors are picked out with dimension reduction and collinear diagnosis to eliminate the useless and redundant data.The processed data are analyzed and verified with examples.

关 键 词：决策树 流失预警 降维 ID3算法 CART算法 LOGISTIC回归 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论]