Prediction of traffic mode selection based on MNL model

The analysis of travel mode choice is important in transportation planning and policy-making in order to understand and forecast travel demands. The Multinomial Logit (MNL) Model is a powerful method for constructing an ensemble of random decision trees. It de-correlates the decision trees in the ensemble via utility random term that leads to an improvement of forecasting and reduces the variance when averaged over the trees. This paper proposes a nonrandom subset Multinomial Logit method to analyze travel mode choices for examining the prediction capability and model interpretability. Using SP survey data of Swiss metro system in 2017, the influence of model parameters on prediction performance is studied. The results show that MNL method has a high prediction accuracy (82.18%) and a low computational amount, and has a good effect in the prediction of travel mode selection. The prediction effect of subway is 83.96%, so the construction of subway should be preferred in urban transportation planning.