Correlation Analysis between Taxi Order Revenue Level and Multi-Source Data Based on PSO-XGBOOST Model

Le résumé fourni par la source

Classifying regional categories according to the taxi order revenue level and conducting correlation analysis with POIs and weather can improve the operational efficiency of taxis and the travel experience of residents. In this work, we take Xi’an as the research area and construct a regular hexagonal map grid in ArcGIS and use the natural break method to divide the hexagonal research areas into three levels of low, medium, and high based on the average order revenue. After correlation analysis of independent variables, we construct a system of influencing factor indicators integrating built environment, weather conditions, and time factor and predict revenue levels of different research areas based on the PSO-XGBOOST model and compare it with the XGBOOST, GS-BOOST, and the PSO-FNN models. The SHAP value is utilized to obtain the key influencing factors of revenue level division, which provides an important reference for taxi operation allocation and urban transportation facility system planning.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.