Accuracy analysis of mobile signaling data in measuring travel indices: Based on the comparison with household travel survey

Abstract

Traditional household travel survey provides the basic data for urban traffic planning and the most authoritative record of urban residents' travel behaviors. With the development of information technology, position data represented by mobile signaling data have shown great potential in the field of temporal-spatial behavior research. However, whether the mobile signaling data can replace the travel survey data is controversial because the travel data accuracy of mobile signaling data is difficult to assess. How credible the identification results are and whether the travel characteristics of residents can be accurately measured remains to be answered. In this study, the travel survey data of Shanghai residents and mobile signaling data with similar dates were used to compare the differences in intensity, time, spatial distribution, purpose, and other characteristics of travels. This study also analyzed the degree of deviation and summarized the causes of errors. The research revealed that the mobile signaling data are more accurate in describing the per capita travel times of the travel population and the travel origin-destination (OD) distribution at the scale above the street, but less accurate in describing the travel time distribution and travel purposes. Our research suggests that the main reasons for the deviation in mobile phone data are the impact of incomplete location records on data quality, the impact of retention time thresholds, spatial thresholds, and home and workplace identification. Due to the above systematic errors, mobile signaling data cannot replace household travel survey data. Nevertheless, the accuracy of mobile signaling data analysis can be improved by sample screening, trip chain data recovery, and improvement of identification methods.