基于贝叶斯估计的双边车货匹配方法研究

Research on the Method of Bilateral Vehicle-Cargo Matching Strategy Based on Bayesian Estimation

摘要: 为合理调度车主货主提升车货匹配平台的匹配效率，提出一种基于贝叶斯估计的车货匹配概率预测模型和受约束的Gale-Shapley双边匹配算法。首先，应用贝叶斯方法将数据集分为先验概率和条件概率，建立匹配指标计算方法；其次，通过贝叶斯估计计算后验概率，即货车对货物匹配概率值和货物对货车匹配概率值，并将匹配概率值进行排序形成偏好序列；再次，考虑匹配的稳定性和公平性将偏好序列作为数据源代入受约束的Gale-Shapley算法进行求解，得出最终的匹配方案；最后，结合算例分析该模型与算法的有效性与可行性。研究结果表明，基于贝叶斯估计的车货匹配模型与未进行预测的单一匹配模型进行对比，其匹配对成功方案数量可以提升11%，同时平均货车对货物的匹配概率值提升5%，平均货物对货车的匹配概率值提升24%。

Abstract: To rationally schedule vehicle owners and cargo owners and enhance the matching efficiency of the vehicle-cargo matching platform, a Bayesian estimation-based probability prediction model for vehicle-cargo matching and a constrained Gale-Shapley bilateral matching algorithm are proposed. Firstly, the Bayesian method is applied to divide the dataset into prior probabilities and conditional probabilities, establishing a matching index calculation method. Secondly, Bayesian estimation is used to calculate the posterior probabilities, i.e., the probabilities of vehicle-to-cargo matching and cargo-to-vehicle matching, and the matching probabilities are ranked to form preference sequences. Thirdly, considering the stability and fairness of the matching, the preference sequences are taken as data sources to solve the constrained Gale-Shapley algorithm, deriving the final matching scheme. Finally, an example analysis is conducted to evaluate the effectiveness and feasibility of the model and algorithm. The research findings indicate that, when compared to a single matching model without prediction, the vehicle-cargo matching model based on Bayesian estimation can increase the number of successful matching pairs by 11%, while also enhancing the average matching probability of vehicles to cargo by 5% and the average matching probability of cargo to vehicles by 24%.