基于Weka的冲击地压预测方法分析

Analysis of Methods to Predict Rockburst Based on Weka

在运用Weka平台对样本训练和预测时,发现用不同的数据挖掘算法,得到不一样的处理结果。以冲击地压危险程度数据作为样本,利用Weka、Excel和Ultra Edit等软件,分别采用支持向量机,决策树和朴素贝叶斯分类器进行训练和预测。从详细的精度,混淆矩阵和节点错误率这3个方面分别比较3种算法,从而得到结论是:贝叶斯分类器的训练和预测效果是最好的,不仅可以提高准确率,还具有一定的研究价值。

In the use of Weka platform for training and prediction of sample,found that getting the different results when using different data mining algorithm.Takes the rockburst hazard degree as the sample data,and make the training and prediction by Support Vector Machine,Decision Tree and Naive Bayes in Weka,Excel and Ultraedit.Comparison of three algorithms from detailed accuracy by class,confusion matrix and node error rate,would conclude that naive Bayes is best to train and predict among three data mining algorithm,which can not only improve the accuracy,but also has certain research value.