基于动态分类器集成选择和GM(2,1) 的组合预测模型
首发时间:2018-08-03
摘要:由于突发性事件预测过程中往往存在研究样本数据少或者数据缺失的情况,基于分类器集成技术,建立DCESM模型弥补数据缺失的不足,再结合灰色系统GM(2,1),建立动态组合预测模型。DCESM 模型能够充分利用样本数据集中所包含的全部信息,选择恰当的分类器子集对测试样本分类,从而形成标准化的预测样本。灰色系统GM(2,1)是一种研究少数据,贫信息的不确定性性问题的新方法,适用于长期预测。以11组统计数据作为训练样本训练该模型,并利用相对误差对模型进行检验。最终选取5个主要指标,采用该预测模型对广东省2015年雷电灾害指数进行实证分析。结果表明:在等级误判率为11%、模型误差小于20%的情况下,模型精度达到了95%,预测结果符合实际情况。
关键词: 安全工程技术科学;预测模型;缺失数据;分类器DCESM模型;灰色系统GM(2 1)
For information in English, please click here
Combination Forecasting Model Based on the Dynamic Classifier Ensemble Selection and GM(2,1)
Abstract:The prediction of less sample data or missing data often exist in the process of emergencies, classifier integration technology based on lack of established model of DCESM to compensate for the missing data, combined with the grey system model of GM(2,1), dynamic combination. The model of DCESM can make full use of all the information contained in the sample data set, and select the appropriate subset of classifiers to classify the test samples, thus forming a standardized prediction sample. Grey system GM(2,1) is a new method to study the uncertainty problem of less data and poor information, which is suitable for long-term prediction. 11 sets of statistical data are used as training samples to train the model, and the relative error is used to test the model. Finally, 5 main indexes are selected, and the forecasting model is used to analyze the lightning disaster index of Guangdong Province in 2015. The results show that the modCombination Forecasting Model Based on the Dynamic Classifier Ensemble Selection and GM(2,1)el accuracy reaches 95% when the misclassification rate is 11% and the model error is less than 20%,rate is 11% and the model error is less than 20%,actual situation.
Keywords: Safety Engineering Sciences prediction model missing data the model of classifierDCESM grey system GM(2,1)
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
基于动态分类器集成选择和GM(2,1) 的组合预测模型
评论
全部评论0/1000