Trip misreporting mining and expansion method for household travel survey

Trip misreporting involves respondents omitting or concealing actual travel activities during the household travel survey (HTS) process. This paper investigated the trip misreporting mining method within the context of HTS by integrating mobile phone signaling data. Additionally, a data-driven approach was proposed to expand the HTS sample by incorporating trip misreporting and travel characteristics, including travel purpose and mode. The first step involved identifying stationary points through mobile phone signaling data and matching them with HTS data to uncover instances of trip misreporting. Subsequently, Point of Interest (POI) data was utilized to identify travel purposes. By integrating travel characteristic indicators from mobile phone signaling data, a Light Gradient Boosting Machine (Light-GBM) was employed for estimating travel modes. Lastly, a novel method for weighted combination expansion sampling was introduced. This method integrates traditional household and person samples with travel purposes and modes. The study revealed that in Suzhou, China, 23% of total trips went unrecorded in HTS due to trip misreporting. The accuracy of identifying travel purpose and estimating travel mode was 86.9% and 84%, respectively. Compared to the traditional method, this approach yielded a more accurate representation of actual travel distance and time. This reduced the distribution error by 7.4% and 6.2%, respectively. Additionally, six days of metro smart card data from weekdays, weekends, and holidays were utilized to validate daily passenger volume after sample expansion, resulting in a daily bias rate below 5%. The findings can effectively uncover the phenomenon of trip misreporting within HTS and provide innovative approaches to expanding the survey data, thereby enabling a more comprehensive analysis of the urban travel patterns.