Research On The Choice Behavior Of Intercity Travel Mode Based On Machine Learning In The Context Of Big Data

As an important infrastructure for urban agglomerations and metropolitan areas,the inter-city comprehensive transportation corridor is the development focus for realizing the strategy of building a powerful country in transportation and accelerating the construction of a modern comprehensive transportation system.The construction of the inter-city comprehensive transportation corridor must be supported by scientific traffic behavior research,and the research on the choice behavior of inter-city travel mode is an important part of traffic behavior.With the development of the Internet of Things,big data and artificial intelligence,mobile phone signaling data,point of interest data,and machine learning methods have been widely used in research in various fields of transportation.The use of big data processing technology and machine learning methods to mine the big data of intercity travel collected by various types of traffic detectors provides a new idea for studying the choice behavior of intercity travel.This article starts from the influencing factors of intercity travel mode choice behavior,combines the characteristics of big data types such as mobile phone signaling and the modeling requirements of the intercity travel mode choice behavior prediction model,and determines the research scope of influencing factors.Focusing on the three aspects of personal attributes,travel attributes and travel mode attributes,the mobile phone signaling big data,geospatial data and point of interest data are preprocessed.And intercity travel target population identification,travel chain identification,travel mode identification and travel purpose recognition algorithm are designed.This work provides support for the analysis of intercity travel characteristics and the predictive modeling of intercity travel mode choice behavior.The analysis results of the intercity travel data from Chengdu to Mianyang in Sichuan Province as an example show that: the intercity travel big data based on mobile phone signaling data can support the extraction of intercity travel feature data,and its data accuracy can meet the requirements of intercity travel mode selection behavior modeling.Secondly,based on the well-processed inter-city travel big data,this paper constructs three inter-city travel mode choice behavior prediction models based on classification and regression tree model,random forest model and XGBoost model.Using four evaluation indicators,confusion matrix,classification accuracy,f1 value,and running time,the applicability of the machine learning method in the behavior prediction research of intercity travel mode selection was verified in the case analysis,and good results were achieved.Finally,using machine learning explanatory methods such as permutation importance,partial dependence plots and SHAP,the XGBoost model with the strongest predictive performance in this study and its example results are used to analyze the influencing factors of intercity travel mode choice.Draw many qualitative conclusions,such as the timeconsuming intercity travel is the strongest factor affecting the choice of intercity travel mode.And it also quantitatively predicts the sharing rate of inter-city travel modes under the crosseffect of age and different characteristics of mobile phone package tariffs.It shows that the machine learning model and its explanatory methods can provide accurate prediction results for intercity travel behavior researchers and explain the mechanism of influencing factors.