A Passenger Satisfaction Prediction Method Based on Driving Time

Public transport is an important means of transportation for urban residents. In China, the government strongly advocates urban residents to travel by public transport. However, the unreasonable allocation of public transport resources and a large number of passengers result in in-bus congestion during the peak period, which will decrease the passenger satisfaction and increase the loss of passengers. Targeting on the issue, this paper provides a passenger satisfaction prediction method based on Driving time (PSP-DT) using Long Short-Term Memory (LSTM) network. First, the bus arrival and departure time are extracted by matching the bus GPS trajectory with the bus station data using the map matching method. Second, the historical driving time is captured from arrival time and departure time, and the future driving time is predicted by LSTM considering weather. Third, passenger satisfaction based on driving time is estimated by the proportion of overload and standard load. By using the 18 working days data of bus No. 14 and a total of 162,628 bus driving data records in Guangzhou city, China, our experimental results show that 1) the prediction model performs well and the average M̅A̅P̅E̅ and R̅M̅S̅E̅ is 5.57% and 11.40. 2) During the morning and evening peak period, the lowest passenger satisfaction of twenty stations is 45.78% and 71.86% separately.