Research on Transportation Mode Selection Based on Machine Learning

The traffic modes selecting is the third stage of the internationally accepted and mature four-stage method, using the multinominal logit model (MNL), but the model is sensitive to the multicollinearity of independent variables, and the residents’ travel attributes are mostly multi-collinear, and the requirements for random utility are high, so the applicability of the model is poor. Compared with the multinominal logit regression model (MNL), the support vector machine (SVM), and random forest (RF) in the machine learning method has outstanding advantages for classification problems, while avoiding the shortcomings of the MNL model. This paper is based on«Passenger Flow Forecast of Urban Rail Transit Network Planning in Baoji City in 2016»research data of residents’ travel trips. By comparing the accuracy of the choice of transportation mode of the three models, the support vector machine and the random forest model are more improved in accuracy than the MNL model.