• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Conference on ITS Telecommunications >Map-reduce for calibrating massive bus trajectory data
【24h】

Map-reduce for calibrating massive bus trajectory data

机译：Map-reduce用于校准大规模公交车轨迹数据

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Accurate bus trajectory data is the basis of many public transportation applications. However, trajectory data sampled by GPS devices contains notable direction errors. We cannot determine the travelling direction of the bus through trajectory data. To address this problem, we utilize k-nearest neighbor algorithm (K-NN) to determine the direction of the bus trajectory. Meanwhile, the voluminous bus trajectory data accumulated daily need to be process efficiently for further data mining. To meet the scalability and performance requirements, in this paper, we use Map-Reduce programming model for trajectory data direction correcting and projecting the bus GPS point to the road link. Particularly, we compare execution time through setting different amount of reduce to express the extent of running time can be affected. Experimental results indicate that the K-NN algorithm improves the accuracy of the direction field in raw bus trajectory data significantly. By comparing the efficiency under different reduce quantities. The result shows that parallel processing framework improves the computational efficiency by a factor of 2 at least, obtaining.
机译：准确的公交车轨迹数据是许多公共交通应用的基础。但是，GPS设备采样的轨迹数据包含明显的方向误差。我们无法通过轨迹数据确定公交车的行驶方向。为了解决这个问题，我们利用k最近邻算法（K-NN）来确定公交车轨迹的方向。同时，需要对每天累积的大量公交车轨迹数据进行有效处理，以进行进一步的数据挖掘。为了满足可伸缩性和性能要求，在本文中，我们使用Map-Reduce编程模型对轨迹数据进行方向校正，并将公交GPS点投影到道路链路上。特别是，我们通过设置不同的reduce数量来比较执行时间，以表示运行时间可能受到影响的程度。实验结果表明，K-NN算法显着提高了原始客车轨迹数据中方向场的精度。通过比较不同减少量下的效率。结果表明，并行处理框架获得的计算效率至少提高了2倍。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号