基于树增强朴素贝叶斯分类器的出租车制动系统安全状态预测

摘要

The malfunction of the braking system is a main cause of the taxis'accidents on the road，therefore，predicting the working condition of taxi's braking system is meaningful for the management and maintenance on the taxis，reducing the casualty and economic losses caused by traffic accidents. This study is based on the database of 335 cases which is extracted from one of the Hefei Motor Vehicles Safety Technology Inspection stations. Based on three basic vehicle parameters-age，brand and weight，this study builds Tree Augmented Naive Bayesian Classifier（TAN）model，Decision Tree（DT）model and K Nearest Neighbors（KNN） model to predict the working condition of taxi's braking system. The results show that the TAN model outperforms the other two models with higher accuracy，sensitivity and specificity，thus with a good performance the proposed TAN model can be employed to pre-dict the working condition of taxi's braking system usefully.%制动系统故障是引发出租车交通事故的主要原因之一,预测出租车制动系统的安全状态对于主管部门维护管理出租车、减少道路交通伤亡和经济损失具有重要意义.论文基于合肥某机动车安全技术检测站提取的335组出租车制动系统检测数据,以品牌、使用年限和整备质量为属性变量,分别构建树增强朴素贝叶斯分类器模、决策树模型、K近邻模型预测出租车制动系统的安全状态.结果表明,树增强朴素贝叶斯分类器模型的预测准确率、灵敏度、特异性均优于决策树模型和K近邻模型,可准确预测出租车制动系统的安全状态.