Forecasting Model Of Commuter’s Daily Activity-trip Time Allocation

With the increasingly serious traffic congestion, traffic accidents, environmentalpollution and other traffic problems, the traditional method of relying solely onincreasing traffic supply is no longer suitable to solve the imbalances of transportationsupply and demand in the rapid development of urban socio-economic. The emphasisof transportation planning and management has been transformed to the effectivemanagement of traffic demand at this stage.Trip behavior analysis is one of the important research areas of transportationplanning and management. Resident trip survey and trip behavior analysis can helpsolve and ease traffic congestion and other traffic problems gradually. The forecastingof trip time is an important part of the trip behavior analysis and directly affectsresidents’ trip mode, trip routes, the number of trips and activity-trip allocation. As aresult, the forecasting of resident trip time that can provide decision-making basis forthe development and implementation of transportation demand management policy isconducive to master the resident trip pattern and resident trip characteristics.Commuting is a major part of urban residents’ trips. The proportion of Beijingresidents commuting trip to total in2005is26.95%. Commuting time allocation is animportant factor affecting other activities and trips of the urban residents. Becausecommuter trip is concentrating in a peak period and a certain area, it becomes theprominent reason of traffic congestion in the early and evening peak period. Therefore,the study of urban residents’ commuting time characteristics and the forecasting ofresidents’ commuting time is of great significance to understand the characteristics ofresident trip pattern and solve urban traffic congestion problems.The study of traditional commuting time in which the study object is single onlyconsider the departure period and trip time. It does not fully taken into accountresidents’ commuting time. This paper will subdivide commuting time into thediscrete moment and the continuous time and divide commuter’s daily activity-tripallocation into a number of time elements including discrete moment and continuoustime. Thus, it can forecast the commuter’s daily activity-trip time allocation byforecasting these time elements.This paper predicts a number of time elements including discrete moment andcontinuous time through discrete model established by Ordered Probit and continuous model established by SVR based on the commuter’s daily activity-trip chain. Thispaper establishes the model of Commuter’s daily activity-trip time allocation systemthrough the combination of the discrete model and continuous model. The discretemoment includes departure time and halfway park arrival time while the continuoustime includes trip time and activity duration. The result shows that the overall fittingaccuracy of the model is relatively high and this model can predict commuter’s dailyactivity-trip time allocation accurately. It is the innovation point that applying themodel of Commuter’s Daily Activity-trip Time allocation through the combining ofthe discrete model and continuous model to predict discrete moment and continuoustime.The results of this paper can be used to predict the daily activity-trip timeallocation of single commuter and evaluate the transportation demand managementpolicies. It helps master the resident trip pattern and it is a effective method ofresident trip behavior analysis. Combing resident trip characteristics and the results ofthe evaluation can establish more reasonable and more effective traffic demandmanagement policies and provide decision-making basis for traffic planning andmanagement.