面向不平衡数据集的SMOTENC-XGBOOST驾驶人交通安全评估模型

为深入挖掘驾驶人因素与交通事故之间的关系，提出一种基于SMOTENC和极端梯度提升(extreme gradient boosting, XGBoost)的驾驶人交通状态优劣分类算法。首先针对交通事故发生与否不平衡的特点，使用SMOTENC算法对数据进行上采样并在采样过程中加入随机扰动，解决了数据不平衡问题。然后使用Embedded算法结合L1正则化，通过模型评估完成对特征子集的选择。最后使用机器学习的方法将XGBoost算法用于执行数据的特征提取和分类过程。实验表明，在对驾驶人的交通状态进行综合评价的任务上，XGBoost模型的准确率为99.85%，相较于随机森林、支持向量机等对照组模型，提升了约1.12%-1.80%。除此之外，使用SMOTENC算法对数据不平衡问题进行处理后，通过混淆矩阵观察到模型对于好坏个体均具备较好的识别能力。

SMOTENC-XGBOOST Driver Traffic Safety Assessment Model for Unbalanced Dataset

In order to deeply explore the relationship between driver factors and traffic accidents, a classification algorithm based on SMOTENC and Extreme Gradient Boosting (XGBoost) was proposed. Firstly, according to the unbalanced characteristics of traffic accidents, SMOTENC algorithm was used to up-sample the data, and random disturbance was added in the sampling process to solve the problem of data imbalance. Then, using Embedded algorithm combined with L1 regularization, the feature subset was selected through model evaluation. Finally, XGBoost algorithm was used to perform feature extraction and classification of data by machine learning method. Experiments show that the accuracy of XGBoost model is 99.85%, which is about 1.12%-1.80% higher than that of random forest and support vector machine. In addition, after the SMOTENC algorithm is used to deal with the data imbalance problem, it is observed through the confusion matrix that the model has good identification ability for good and bad individuals.