基于Weka的软件缺陷预测研究与应用  

Research and Application of Software Defect Prediction Based on Weka

作　　者：郭江峰 曲豫宾 GUO Jiangfeng;QU Yubin(Jiangsu College of Engineering and Technology,Nantong Jiangsu 226007;Kizilsu Vocational Technical College,Atushi Xinjiang 845350)

机构地区：[1]江苏工程职业技术学院,江苏南通226007 [2]克孜勒苏职业技术学院,新疆阿图什845350

出　　处：《河南科技》2020年第8期14-18,共5页Henan Science and Technology

基　　金：南通市科技计划指令性项目“即时软件缺陷预测模型构建和优化关键技术研究”(JC2018134)。

摘　　要：在软件开发过程中,软件缺陷预测能预先识别存在的潜在缺陷模块,大幅减少测试所需的人力、物力,优化测试资源分配,提高测试效率和软件产品质量。软件缺陷预测技术不仅具有重要的研究意义,更具有重要的应用价值。基于Weka数据挖掘平台,本研究使用NASA缺陷数据集进行了软件缺陷预测,在合理选择机器学习算法、科学设置参数的情况下,取得了良好的软件缺陷预测结果。In the process of software development, software defect prediction can identify the potential defect modules in advance, greatly reduce the human and material resources needed for testing, optimize the distribution of testing resources, and improve the testing efficiency and software product quality. Software defect prediction technology not only has important research significance, but also has important application value. Based on Weka data mining platform, the software defect prediction was carried out by using NASA defect data set in this study, and good software defect prediction results were obtained under the condition of selecting machine learning algorithm reasonably and setting parameters scientifically.

关 键 词：软件缺陷预测 度量元 机器学习 WEKA 

分 类 号：TP311.52[自动化与计算机技术—计算机软件与理论]