一种基于SVD的交通流量数据补全算法

A Traffic Flow Data Completion Algorithm Based on SVD

针对高速公路流量数据采集容易出现数据缺失的情况,提出一种基于奇异值分解(SVD)的数据补全算法。首先对数据进行处理,再对其进行数据分解得到特征量,利用特征量找出相似日期,最后利用相似日期进行缺失数据估计,从而补全数据。与传统的均值补全算法进行了相比分析,本算法具有较好的性能。

In view of the highway traffic flow data collection prone to absence of data,a data completion algorithm is proposed. First, the history data is processed, then decompose the data to get the feature vectors, find the similar dates using the feature vectors, at last, using similar dates ＇ data to estimate missing data and complete data. Compared with traditional mean completion algorithm, this algorithm has good performance.