Car use model based on tour mode and activity choice interdependencies

To improve the conventional car travel forecasting method, an active-based car use model has been constructed. Firstly, the activity pattern of car user was statistically analyzed, and the car use patterns was classified, then a co-evolutionary Logit car use prediction model was built based on MNL (multinomial Logit) models. The co-evolutionary method can present the interdependence between tour mode and activity choice, and emulate the decision order of tour mode and activity choice. The adaptive analysis proves that the model can be used for travel demand forecasting and transportation demand management policy evaluation. The data from Shenyang city is taken as an example in this paper. Results indicate that 88.9% forecasting results of the new model are correct, in 65% cases the activity decisions are made after the car use choice. The model fully verifies the variation of tour mode and activity choice decision order, thus the accuracy of car travel prediction is improved.