基于机器学习的换乘优惠政策对不同收入水平乘客的实施效果评估

Effect Evaluation of Transfer Preferential Policy on Passengers with Various Income Based on Machine Learning

换乘优惠政策是促进城市居民低碳出行的重要手段,评估换乘优惠政策实施效果对城市公共交通规划与管理具有重要意义。该文建立了一种基于卷积神经网络与随机森林的精细城市房价模拟模型,并将其模拟结果与换乘优惠政策实施前后的乘客IC刷卡数据耦合,实现对各收入水平乘客的准确划分,在此基础上通过优惠享受率、客流特征以及核密度分析等评估城市公共交通换乘优惠政策的实施效果。对2019年厦门市换乘优惠政策案例的研究结果表明:厦门市房价(表征居民收入水平)分布呈岛内高值集中、岛外低值均匀分布的特点,换乘优惠政策对中低收入水平乘客的吸引效果最强;当乘客出行总花费优惠至4元左右时,接驳客流吸引效果达到最优。研究结果可为今后城市公共交通换乘优惠政策制定与优化提供参考。

Transfer preferential policy(TPP)is an important means to promote low-carbon travel of urban residents.Comprehensive and detailed evaluation of the implementation effect of the transfer preferential policy is of great significance to urban public transport planning and management.Housing price is an important indicator of residents′income,and passengers with various income have different travel characteristics,so the implementation effect of transfer preferential policies for passengers with various income is different.This paper establishes a fine-scale urban housing price prediction model based on convolutional neural network and random forest,and couples its prediction results with passenger IC card swiping data before and after the implementation of the policy,so as to realize the accurate division of passengers by income.Finally,the implementation effect of TPP is evaluated through the preferential rate,passenger flow analysis and kernel density analysis,etc.The case study of preferential transfer policies in Xiamen in 2019 shows that:for the distribution of housing prices in Xiamen,the high values are concentrated in the island and the low values are evenly distributed outside the island;the policy has the strongest effect on attracting passengers with low-middle income;the optimal flow-attractive effect can be achieved when the total travel cost of passengers is preferential to about 4 yuan.The results of this paper can provide a basis for the formulation and optimization of TPP in the future.