Destination Estimation of Passenger Trip Based on Smart Card Data

• Main Authors: LU, CHI-HUA, 呂奇樺
• Other Authors: YEH, CHAO-FU
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/289b73

碩士 === 逢甲大學 === 運輸科技與管理學系 === 105 === Smart card data have been used in different domestic and foreign research areas because smart card data are more precise and all-round data compared with traditional travel survey. Smart card data contain a lot of useful information, e.g. date, time, bus stop, boarding and alighting stop/time, ID number of smart card and so on. Because of the advantages of the smart card, both of urban bus and intercity bus are built up the contactless smart card system in Taiwan. Although smart card data can offer useful information, it’s still limited to how the passengers are being charged, e.g. mileage charges or pay-per-segment. If the bus fare is charged through the mileage charges, the smart card database can collect both of the boarding and alighting stop of the passenger. If the bus fare is charged through pay-per-segment, the smart card database can only collect the boarding or the alighting stop of the passenger. Pay-per-segment cause an incomplete data, so the data can’t be used for the transportation planning directly. In order to solve this issue, this paper focuses on estimation of the destination and presents an algorithm to estimate the destination location for each individual boarding of the bus with a smart card. Drawing on Taichung city bus as the case study, the evaluating result of accuracy for predicting algorithm is 63 percent with 600 meters of tolerance distance. After considering the tolerance distance error, the accuracy for predicting algorithm is increased to 90 percent. This algorithm can more be used in other areas where have incomplete smart card data in order to improve the planning of public transport network.