基于改进KNN算法的交通流异常数据修复方法

交通流数据分析是交通规划、控制、管理等工作实施的基础。交通流数据异常会给交通状态辨识及交通管理和控制带来困扰，不利于交通领域各方面研究及工作的开展。因此，对异常数据进行修复具有必要性。为了提高交通流异常数据修复精度，进一步改善交通数据质量，构建了基于改进K近邻（K-Nearest Neighbor，KNN）算法的交通流异常数据修复模型。通过对KNN基础模型中k值和状态向量进行优选、提出距离占比的近邻值权重选取方式，实现对其模型的改进。为了验证模型的有效性，采用实测交通流数据进行实验分析。实验结果表明，改进的KNN数据修复模型具有更高的修复精度，其平均相对误差为9.88%，能够有效改善数据质量，为智能交通控制体系提供基础数据支持。

A recovery method for abnormal traffic flow data based on the improved KNN algorithm

Traffic flow data analysis is the basis for implementation of traffic planning, control, and management. Abnormal traffic data which is not conducive to all aspects of transport research and related work brings difficulties to the identification of traffic conditions, traffic management and control. Therefore, it is necessary to repair abnormal data. To improve the recovery accuracy of traffic flow anomaly data and further improve the quality of traffic data, a model of traffic flow anomaly data recovery method based on improved K-Nearest Neighbor(KNN) algorithm was constructed. The model is improved by optimizing the k-values and state vectors in the KNN basic model, and proposing a distance weights as the selection of neighboring weights values. In order to verify the validity of the model, the measured traffic flow data was used for experimental analysis. The experimental results show that the improved KNN data recovery model has higher recovery accuracy, the mean average relative error is 9.88%. It can effectively improve the data quality, and provide basic data support for the intelligent traffic control system.