数据挖掘在预测组织事故防控效果中的应用*

为进一步探索数据挖据技术在组织事故预防工作中的融入性与适用性，基于24Model构建事故预控基础模型,通过预测准确率数值及接受者操作特性曲线（ROC曲线）对比分析随机森林（RF）、支持向量机(SVM)、决策树（DT）与神经网络（NN）4种方法对组织事故防控效果的预测性能。结果表明:针对事故率控制（Y1）、职业危害预防（Y2）、财产损失3类预测目标（Y3），RF方法均能达到较高的准确率及稳定性，具有较优的预测性能。根据特征重要度（FI）排序，明确对组织事故水平影响最显著的因素为安全实践活动认知（SC5）及安全管理程序文件（SMS3），FI值均大于0.150 0。研究结果可为有效预测组织事故防控效果提供方法依据，同时为企业安全工作的规划设计提供思路。

Application of data mining in predicting prevention and control effect of organizational accidents

In order to further explore the integration and applicability of data mining technology in the prevention of organizational accidents,a basic model of accident pre control was constructed according to 24Model.The prediction effect of four methods,including random forest (RF),support vector machine (SVM),decision tree (DT) and neural network (NN),were compared and analyzed based on the values of prediction accuracy and the receiver operating characteristic (ROC) curve.The results showed that aiming at three types of prediction target including the accident rate control（Y1）,occupational hazard prevention（Y2）and property loss（Y3）,RF method could all achieve the higher accuracy and stability with the better prediction performance.According to the rank of feature importance (FI) ,it was clarified that the factors with the most significant influence on the organizational accident level were the safety practice awareness (SC5) and safety management procedure documents (SMS3),with both the values of FI larger than 0.150 0.The research provides the method basis to effectively predict the prevention and control effect of organizational accidents,as well as the ideas for the planning and design of safety work in the enterprises.