| 张翔宇,张强,吕明琪.基于演化模式挖掘和代价敏感学习的交通拥堵指数预测[J].高技术通讯(中文),2020,30(9):918~927 |
| 基于演化模式挖掘和代价敏感学习的交通拥堵指数预测 |
|
| |
| DOI：doi：10.3772/j.issn.1002-0470.2020.09.006 |
| 中文关键词: 交通拥堵指数预测； 序列模式挖掘； 代价敏感学习； 数据融合； 城市计算 |
| 英文关键词: traffic congestion index prediction, sequential pattern mining, cost-sensitive learning, data fusion, urban computing |
| 基金项目: |
|
| 摘要点击次数: 2120 |
| 全文下载次数: 1414 |
| 中文摘要: |
| 交通拥堵指数预测是智能交通系统的核心能力之一。然而，现有方法大多采用回归模型，在长期交通拥堵指数预测任务上表现不佳。针对此问题，本文提出了一种融合演化模式挖掘和代价敏感学习的交通拥堵指数预测方法。首先，采用序列模式挖掘算法从交通拥堵指数历史数据中发现长期演化模式。同时，采用代价敏感学习技术对交通拥堵指数数据与多种时空特征之间的关联进行学习。最后，通过Stacking框架对演化模式挖掘和代价敏感学习的能力进行融合。基于杭州市真实交通拥堵指数数据集进行的实验表明，本文提出的方法对未来5天交通拥堵指数的预测误差比现有方法降低了10%以上。 |
| 英文摘要: |
| Traffic congestion index prediction is one of the key ability of intelligent transportation system. However, the existing methods mostly apply regression techniques, resulting in poor performance on long-term traffic congestion index prediction.Aiming at this problem, this paper proposes a hybrid traffic congestion index prediction method by fusing sequential pattern mining and cost-sensitive learning. First, it discovers long-term evolving patterns from the historical traffic congestion index data by using sequential pattern mining algorithm. Second, it learns the correlations between traffic congestion index data and a variety of spatiotemporal features by using cost-sensitive learning technique. Finally, it fuses the ability of sequential pattern mining and cost-sensitive learning based on Stacking framework. The method is evaluated based on real datasets from Hangzhou city, and the experiment results show that the proposed method reduces the prediction error by over 10% compared to the state-of-the-art methods. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
