| 短时交通流预测的改进K近邻算法 |
| |
| 引用本文： | 谢海红, 戴许昊, 齐远. 短时交通流预测的改进K近邻算法[J]. 交通运输工程学报, 2014, 14(3): 87-94. |
| |
| 作者姓名： | 谢海红 戴许昊 齐远 |
| |
| 作者单位： | 1.北京交通大学 城市交通复杂系统理论与技术教育部重点实验室, 北京 100044;;2.北京城市交通协同创新中心, 北京 100044;;3.湖南省交通规划勘察设计院, 湖南 长沙 410008 |
| |
| 基金项目： | 国家973计划项目2012CB725403 |
| |
| 摘 要： | 分析了原有的短时交通流预测的K近邻算法, 用模式距离搜索方法代替原有的欧氏距离搜索方法, 引入多元统计回归模型, 建立了一种改进的短时交通流预测的K近邻算法, 并以北京市某路段进行实例验证。试验结果表明: 当K取23时, 利用改进的K近邻算法, 预测结果的均方误差、平均相对误差、平均绝对误差分别为31.43%、4.17%、0.27%;利用原有的K近邻算法, 预测结果的均方误差、平均相对误差、平均绝对误差分别为33.33%、4.40%、0.28%;利用历史平均模型, 预测结果的均方误差、平均相对误差、平均绝对误差分别为46.20%、11.40%、0.48%。可见, 改进的K近邻算法的预测精度明显高于其他2种方法, 在提高搜索效率的同时准确地刻画了交通流的真实情况。
|
| 关 键 词： | 交通规划 短时交通流预测 K近邻算法 模式距离 多元统计回归 |
| 收稿时间： | 2014-01-13 |
Improved K-nearest neighbor algorithm for short-term traffic flow forecasting |
| |
| XIE Hai-hong, DAI Xu-hao, QI Yuan. Improved K-nearest neighbor algorithm for short-term traffic flow forecasting[J]. Journal of Traffic and Transportation Engineering, 2014, 14(3): 87-94. |
| |
| Authors: | XIE Hai-hong DAI Xu-hao QI Yuan |
| |
| Affiliation: | 1. MOE Key Laboratory for Urban Transportation Complex Systems Theory and Technology, Beijing Jiaotong University, Beijing 100044, China;;2. Center of Cooperative Innovation for Beijing Metropolitan Transportation, Beijing 100044, China;;3. Hunan Provincial Communications Planning, Survey and Design Institute, Changsha 410008, Hunan, China |
| |
| Abstract: | The original K-nearest neighbor algorithm for short-term traffic flow forecasting was analyzed.Pattern distance search method was used to replace the original Euclidean distance search method, the multiple statistics regression model was introduced, an improved K-nearest neighbor algorithm for short-term traffic flow forecasting was put forward, and an example verification was carried out by using the traffic flow data from a certain section in Beijing.Test result indicates when Kis 23, the error of mean square, mean absolute error and average relative error of forecasting results are 31.43%, 4.17% and 0.27% respectively by using the improved K-nearest neighbor algorithm.By using the original K-nearest neighbor algorithm, the error of mean square, mean absolute error and average relative error of forecasting results are 33.33%, 4.40% and 0.28% respectively.By using the historical average model, the error of mean square, mean absolute error and average relative error of forecasting results are 46.20%, 11.40% and 0.48% respectively.The forecasting accuracy of the improved K-nearest neighbor algorithm is obviously higher than the other two algorithms.The improved K-nearest neighbor algorithm notonly increases searching efficiency, but also accurately reflects the real situation of traffic flow. |
| |
| Keywords: | traffic planning short-term traffic flow forecasting K-nearest neighbor algorithm pattern distance multiple statistical regression |
|
| 点击此处可从《交通运输工程学报》浏览原始摘要信息 |
|
点击此处可从《交通运输工程学报》下载全文 |
|
