[1]胡燕,朱晓瑛,马刚.基于K-Means和时间匹配的位置预测模型[J].郑州大学学报(工学版),2017,38(02):17-20.[doi:10.13705/j.issn.1671-6833.2017.02.005]
　Hu Yan,Zhu Xiaoying,Ma Gang.Location Prediction Model Based on K-Means Algorithm and Time Matching[J].Journal of Zhengzhou University (Engineering Science),2017,38(02):17-20.[doi:10.13705/j.issn.1671-6833.2017.02.005]
点击复制

基于K-Means和时间匹配的位置预测模型()
分享到：

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
38
期数:
2017年02期
页码:
17-20
栏目:
出版日期:
2017-04-28

文章信息/Info

Title:
Location Prediction Model Based on K-Means Algorithm and Time Matching
作者:
胡燕; 朱晓瑛; 马刚
北京邮电大学信息网络中心,北京,100876
Author(s):
Hu Yan;  Zhu Xiaoying;  Ma Gang
Information Network Center of Beijing University of Posts and Telecommunications, Beijing, 100876 
关键词:
位置预测; K-Means算法; 时间匹配; 聚类
Keywords:
location prediction; K-Means algorithm; time matching; cluster
DOI:
10.13705/j.issn.1671-6833.2017.02.005
文献标志码:
A
摘要:
随着移动服务的发展,越来越多的移动端服务基于对象的位置进行推送和推荐,因此位置预测技术显得越来越重要.由于对象位置信息存在采集不连续或对象行为不规律等因素,导致位置预测成为一项非常有挑战的工作.为了提高位置预测的准确性,提出一种基于K-Means算法和时间匹配的位置预测模型.该模型使用K-Means算法对历史位置点进行聚类,划分多个对象运动区域,针对对象运动区域进行预测.按照对象的作息时间将一天时间划分为多个时间段,运用笔者提出的轨迹建模算法和轨迹更新算法形成用户运动轨迹,形成对象运动轨迹,再使用时间匹配原则进行位置预测.笔者最后利用真实的数据实现该模型,实验证明:未使用该模型的位置预测准确率为39.7％;使用该模型后算法和时间匹配的位置预测模型预测准确率达到60.3％,准确率提高了20％左右.
Abstract:
Location prediction was critical to mobile service because various kinds of applications were tightly combined with user’s location.However,location prediction was a challenging work because location capturing was always not continuous and user’ s behavior were uncertain and irregular.To improve the location prediction accuracy rate,this paper proposed a location prediction model based on K-Means algorithm and time matching.For the mobile service always region oriented,we first clusted history location using K-Means algorithm to define several regions.Then we divided every day time into several segments and calculated the maximum probability location in every time segment.A trajectory of a user in one day was formed with trajectory model and trajectory updating model which proposed in this paper.We could predict user’ location with time matching method.At last,we did experiments with real location data in campus which captured by APs.The prediction out come with K-Means was compared to the outcome without model based on K-Means algorithm.The experiment result shows that accuracy rate of our model was higher than the prediction without new model.So,more location services could be provided to users with this new model.

相似文献/References:

[1]段炼,党兰学,李铭,等.位置数据稀疏约束下的疑犯时空位置预测[J].郑州大学学报(工学版),2018,39(05):58.[doi:10.13705/j.issn.1671-6833.2018.05.003]
　Duan Lian,Dang Lanxue,Li Ming,et al.Spatiotemporal Prediction of Suspect under Location Data Sparsity Constraint[J].Journal of Zhengzhou University (Engineering Science),2018,39(02):58.[doi:10.13705/j.issn.1671-6833.2018.05.003]
[2]张忠林,曹志宇,李元韬.基于加权欧式距离的k-means算法研究[J].郑州大学学报(工学版),2010,31(01):89.
　ZHANG Zhonglin,CAO Zhiyu,LI Yuantao.Research Based on Euclid Distance with Weights of K——means Algorithm[J].Journal of Zhengzhou University (Engineering Science),2010,31(02):89.

更新日期/Last Update: