摘要
为了提高软件缺陷预测的准确率,利用布谷鸟搜索(cuckoo search,CS)算法的寻优能力和人工神经网络(artificial neural network,ANN)算法的非线性计算能力,提出了基于CS-ANN的软件缺陷预测方法。此方法首先使用基于关联规则的特征选择算法降低数据的维度,去除了噪声属性;然后利用布谷鸟搜索算法寻找神经网络算法的权值,使用权值和神经网络算法构建出预测模型;最后使用此模型完成缺陷预测。使用公开的NASA数据集进行仿真实验,结果表明该模型降低了误报率,并提高了预测的准确率,综合评价指标AUC(area under the ROC curve)、F1值和G-mean都优于现有模型。
To improve the accuracy of software defect prediction, this paper proposed a software defect prediction method based on CS-ANN, which took advantage of parameters optimization power of the cuckoo search (CS) and non-linear computing power of the artificial neural network ( ANN ). The method firstly used the feature selection algorithm based on the association rules to reduce the dimension of the original data sets and remove the noise attributes existing in data, and found the weights of the neural network by using the CS algorithm. Then the method built the defect prediction model using the weights and neural network algorithm. Finally, this paper completed defect prediction by the model. The simulation experiment was performed by using the NASA datasets that were published. The results show that this model can reduce the false alarm rate and improve the prediction accuracy, moreover, comprehensive evaluation area under the ROC curve (AUC), Fl-measure and G-mean values are superior to existing models.
出处
《计算机应用研究》
CSCD
北大核心
2017年第2期467-472,476,共7页
Application Research of Computers
基金
国家自然科学基金资助项目(61379032,61462091,61262025)
云南省教育厅科学研究基金资助项目(2015Z018)
云南大学博士科研启动项目(XT412004)
关键词
软件缺陷预测
人工神经网络
布谷鸟搜索算法
软件质量
机器学习
software defect prediction
artificial neural network
cuckoo search
software quality
machine learnin