南京理工大学学报（自然科学版）2017，Vol.41Issue(4)：448-453,6.DOI:10.14177/j.cnki.32-1397n.2017.41.04.008
一种基于改进最近邻算法的忠诚度预测方法
Loyalty prediction method based on improvednearest neighbor algorithm
摘要
Abstract
In order to improve the accuracy and efficiency of loyalty prediction,this paper,combining K nearest neighbor(KNN)with Bayesian algorithm as a classification prediction method,proposes a loyalty prediction method based on improved nearest neighbor.The method first takes high loyal customers and low loyal customers as the same category of loyal customers.The method classifies the data set with Bayesian algorithms to obtain non-loyal customers and loyal customers.It takes loyal customers as test data set of the following KNN algorithm and classifies them to obtain high loyal customers,low loyal customers and non-loyal customers.The experimental results show that this method can not only reduce the impact of the K value on the nearest neighbor algorithm and reduce its memory overhead,but also can effectively shorten the time of loyalty classification and improve the accuracy of the classification accuracy.关键词
数据挖掘/分类/聚类/回归/K最近邻算法/贝叶斯算法/忠诚度预测Key words
data mining/classification/clustering/regression/nearest neighbor algorithm/bayesian algorithm/loyalty prediction分类
信息技术与安全科学引用本文复制引用
朱虹,李千目,戚湧..一种基于改进最近邻算法的忠诚度预测方法[J].南京理工大学学报（自然科学版）,2017,41(4):448-453,6.基金项目
国家重点研发计划政府间国际科技创新合作重点专项(S2016G9070) （S2016G9070）
江苏省重大研发计划社会发展项目(BE2017739) （BE2017739）
江苏省重大研发计划产业前瞻项目(BE2017100) （BE2017100）
中央高校基本科研业务费专项资金(30916015104) （30916015104）
赛尔下一代互联网创新项目(NGII20160122) （NGII20160122）
中兴通讯产学研合作论坛合作项目(2016ZTE04-11) （2016ZTE04-11）
