基于WEKA的软件维护性评价模型的实验研究  

Experimental research in evaluation model of software maintainability based on WEKA

作　　者：刘丽[1,2] 朱小冬[1,3] 叶飞[1] 

机构地区：[1]军械工程学院装备指挥与管理系,河北石家庄050051 [2]石家庄机械化步兵学院教研部,河北石家庄050003 [3]武汉大学计算机学院,湖北武汉430072

出　　处：《河北科技大学学报》2012年第1期69-73,共5页Journal of Hebei University of Science and Technology

基　　金：装备预先研究项目

摘　　要：为了能够对软件维护性进行评价,以类级软件维护性数据集为例进行了软件维护性实验,应用Logiscope对类的维护性做了定性评价;运用Krakatau Professional收集类各个度量的度量值;通过WEKA得到了类级度量关键属性;最后,选取决策树分类方法训练分类模型,并利用训练好的分类模型对测试集进行测试,得到基本满足要求的类级软件维护性评价模型。In order to evaluate software maintainability,an experiment was made about class dates of software maintainability. Class maintainability was qualitatively evaluated by applying Losiscope.Some class metrics were collected by using Krakatau Professional,metric attributes were analysed through WEKA and the key attributes were obtained.At last,professional class model was trained through decision tree classifier,and the model was tested.

关 键 词：软件维护性评价 LOGISCOPE Krakatau PROFESSIONAL WEKA 决策树分类器 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]