为了解决城市区域火灾风险预警,以消防大数据资源为基础利用数据挖掘技术建立火灾预警模型.利用潍坊区域火灾信息和环境因素相关信息,采用多种数据挖掘方法对火灾风险统计数据进行模型分析和比较.研究结果表明,在使用支持向量机(SVM)模型以及温度、相对湿度、雨量、风力四项环境属性时模型具有最好的预测效果.基于数据扰动对于结果的影响程度设计算法,分别得到这四种属性对于预测结果重要性的权重值,该结论可用于同类区域构建预测模型时的数据预处理.%In order to solve the fire risk alerted problem in urban ar-ea, a fire risk alerted model was set up by using data mining tech-nology on the basis of large fire data resources. Using the data of fire and environment information in Weifang, several data mining methods were used to model and compare the fire risk statistical da-ta. Research showed that, the model has the best fitting effect when using support vector machines (SVM) and four environmen-tal attributes (temperature, relative humidity, rainfall, wind force). Design a calculation method based on the influence of data perturbation on the results, obtained the weights of the importance of the four attributes to the prediction results. The results can be used for the data pretreatment of similar area prediction model building.
展开▼