摘要
滑坡灾害预测受多种因素影响,其中降雨等不确定因素存在难以获取数据及有效处理等难题,为提高滑坡危险性预测的准确率,根据滑坡灾害发生相关理论及决策树分类原理,提出了基于不确定决策树算法在滑坡危险性预测的方法。该方法引入不确定因子降雨,并将不确定因子和其余评价因子一起,根据不确定决策树算法理论构建出不确定决策树,建立滑坡危险性等级预测模型,并用延安市宝塔区的实例进行验证。实验结果表明,该预测方法取得了较高的总体精度和有效精度,达到了滑坡预测的精度标准,且两项预测精度均高于传统C4.5决策树方法。
The prediction of landslide hazard is affected by many kinds of factors. In the process of prediction,there are some difficulties in acquiring data of uncertain attribute rain and dealing with these data effectively. Aiming at improving the forecast accuracy of landslide hazard,this paper put a landslide hazard assessment based on uncertain decision tree classification method forward on the basis of correlation theory of landslide hazard and decision tree classification theory. This method introduced uncertain factor rain and put it together with other assess factors to build an uncertain decision tree according to uncertain decision tree classification theory. Then it built a landslide hazard level prediction model,selected Baota district of Yanan city as the study area to test the accuracy of this model. The test result shows that the uncertain decision tree classification method can get a high total accuracy and effective accuracy and meet the accuracy standard of landslide hazard prediction. By using this method,both the total accuracy and effective accuracy of landslide hazard prediction are higher than traditional C4. 5 decision tree method.
出处
《计算机应用研究》
CSCD
北大核心
2014年第12期3646-3650,共5页
Application Research of Computers
基金
国家"863"计划资助项目(2012AA061901)
国家自然科学基金资助项目(41362015
51164012)
江西省自然科学基金资助项目(20122BAB201045)
国土资源调查项目(1212011140005)
关键词
不确定数据
决策树
危险性预测
uncertain data
decision tree
hazard level prediction