基于XGBoost算法的道路交通事故严重程度预测  

Prediction of Road Traffic Accident Severity Based on XGBoost Algorithm

作　　者：王延安 刘庆芳 成卫[2] WANG Yan-an;LIU Qing-fang;CHENG Wei(Traffic Police Detachment of Yuxi Public Security Bureau,Yuxi 653100,China;School of Traffic Engineering,Kunming University of Science and Technology,Kunming 650000,China)

机构地区：[1]玉溪市公安局交通警察支队,云南玉溪653100 [2]昆明理工大学交通工程学院,云南昆明650000

出　　处：《软件导刊》2022年第5期84-88,共5页Software Guide

摘　　要：为了进一步加强道路交通安全管理,提升道路交通安全预警系统的准确度,提出一种基于XGBoost算法的道路交通事故严重程度预测模型。首先利用SMOTE对不平衡数据集进行处理,使正负样本数量达到1∶1;然后利用随机森林算法将影响城市道路交通事故严重程度的特征进行重要性排序,找出对预测模型影响较大的因素;最后基于XGBoost算法构建预测模型,使用网格搜索法进行模型参数寻优,提高预测准确度。通过与KNN、Logistic及随机森林3种模型进行结果对比分析发现,XGBoost模型的分类准确率平均提升0.097。基于XGBoost算法的道路交通事故严重程度预测模型拥有更加优越的预测性能,可为预防和降低交通事故严重程度提供可靠参考。In order to further strengthen road traffic safety management and improve the accuracy of road traffic safety early warning system,a road traffic accident severity prediction model based on XGBoost algorithm is proposed.First,SMOTE was used to process the unbalanced data set,and the number of positive and negative samples reached 1∶1.Then the random forest algorithm was used to rank the characteristics of urban road traffic accident severity in order of importance to find out the factors that had a greater impact on the prediction model.Finally,the prediction model is built based on XGBoost algorithm,and the grid search method is used to optimize the model parameters to improve the prediction accuracy.By comparing the results with KNN,Logistic and random forest models,the classification accuracy of XGBoost model was improved by 0.097 on average.The road traffic accident severity prediction model based on XGBoost algorithm has better prediction performance,which can provide a reliable reference for preventing and reducing the severity of traffic accidents.

关 键 词：交通安全 事故严重程度 SMOTE 随机森林 XGBoost 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]