基于完全二次型分类器对机场出租车司机决策行为的研究

Research on Decision-making Behavior of Airport Taxi Drivers Based on the Complete Quadratic Classifier

为了探究机场出租车司机合理的决策行为,文章搜集了上海虹桥机场附近的出租车轨迹数据,依据自定义的司机等待时间成本和空载时间成本,并结合其他因素,构建影响出租车司机决策的影响因素体系。然后通过灰色关联度分析从体系中选取具有说服力的影响因素,即灰色关联度数值大于0.65的因素,最后利用所建立的完全二次的Logistic二分类器模型对机场不同出租车司机的决策行为进行预测,结果表明完全二次的二分类器在数据集上能实现96.77%的预测准确率。

In order to determine the decision-making behavior of airport taxi drivers.The article starts to clean data,according to the custom driver waiting time cost and the idle time cost,we established the index system of influence the decision-making behavior of the taxi driver in combination with other factors,then through the grey correlation analysis to select persuasive influence factors from system,which means the influence factors of gray correlation degree value is greater than 0.65.Finally,we build completely quadratic Logistic two classifiers to forecast airport taxi driver of different decision-making behavior,and find that established model in data sets can achieve 96.77%forecast accuracy.