论文部分内容阅读
交通拥挤识别实质上是一种不平衡分类问题,通过解决不平衡分类问题,在数据层面对原始数据集进行重采样,并采用不同的采样倍率进行向上和向下采样,降低数据集类间不平衡程度,从而提高拥挤类识别精度。选取南京市虎踞路(主干道)某一路段作为研究对象,调查获得7:30—9:00交通流数据,并在此基础上,通过vissim软件仿真得到更多数据。借助weka软件平台运用朴素贝叶斯分类器进行分类试验,并对检测结果对比分析,结果表明重采样方法在对总体识别率影响较小的情况下,能够提高拥挤类的识别率。
Traffic congestion recognition is essentially an unbalanced classification problem. By solving the problem of unbalanced classification, the original data set is resampled at the data level, and different sampling magnifications are used to sample up and down to reduce the difference between data sets Balance degree, thereby improving the accuracy of crowding class identification. A certain section of Huju Road (main trunk road) in Nanjing City was selected as the research object. Traffic flow data of 7: 30-9: 00 were obtained and based on this, more data were obtained through software simulation of vissim. Weka software platform using Naive Bayesian classifier classification test, and the test results were compared and analyzed, the results show that the resampling method in the case of less impact on the overall recognition rate, can increase the recognition rate of crowded class.