Study on the Floating-Car Data Pretreatment Approach
Zhenjun Wang, Guang-sheng Jiang, Jing Liu, Jizhen Guan · 2010
Floating cars provide a cost-effective and reliable way to collect traffic data and are particularly suitable for large cities. There are about 10,000 taxies equipped with Global Positioning Satellite (GPS) systems in Beijing, China. Based on the research of traffic flow characteristics of Floating-Car Data (FCD), this paper analyzes the methods for improving the data quality of floating cars in traffic flow time series and studies the FCD pretreatment approach, which includes the discrimination of deficient data, the criteria of abnormal fluctuations, the determination of the minimum sample size, the method of data de-noising, and data recovery. Reliability of the road network and accuracy of short-time travel time predictions can be greatly improved by floating-car original data being pretreated.