无车承运人平台线路定价模型研究

Research on route pricing model of car-free carrier platform

为实现承运线路的科学定价,首先,基于历史定价数据,进行斯皮尔曼相关性的分析,得到变量之间的相关性系数.使用逐步回归法来对变量的显著性进行检验从而剔除掉不显著的变量,使用Rlasso算法对变量进行进一步的筛选,得到最终12个相关变量.并使用了R聚类模型再次对定量变量进行聚类验证,将变量分为4类,并选取了4个代表变量,使用专家打分-Topsis算法进行评价.基于3种BP神经网络预测模型,首先对前期的数据进行训练和测试,从而选取了最优的Levenberg-Marquardt-BP模型.使用最优模型对接下来问题的报价及线路成本以及交易时长建立了预测模型.最终,提出了调整定价具体的措施和建议的算法,为无车承运人平台的定价问题提供参考.

Since the opening of the domestic road transportation market, the development status of small, scatted and chaotic has gradually formed. With the gradual rise of China’s car-free carrier industry, the scientific pricing of carrier routes is an urgent problem of many car-free carrier platforms. First, based on past historical pricing data, we perform Spearman correlation analysis to obtain the correlation coefficient between variables. Stepwise regression is used to test the significance of the variables to eliminate insignificant variables, and the Rlasso algorithm is used to further filter the variables to obtain the final 12 relevant variables. the R clustering model was used to cluster the quantitative variables again, the variables were divided into four categories, and four representative variables were selected, and the expert scoring-Topsis algorithm was used to make evaluation. Based on three BP neural network prediction models, firstly, the previous data was trained and tested, and the optimal Levenberg-Marquardt-BP model was selected. The optimal model is used to establish a predictive model for the quotation, line cost, and transaction duration of the next problem. Finally, based on the previous research, we put forward specific measures and suggested algorithms for adjusting pricing, which provide a reference to the pricing issues of car-free carrier platforms.