基于改进蛇优化算法的软件缺陷预测方法研究  

Software defect prediction method based on improved snake optimization algorithm

作　　者：刘卫光 杨苏琦 张文宁[1] 李学相[3] LIU Weiguang;YANG Suqi;ZHANG Wenning;LI Xuexiang(School of Software,Zhongyuan University of Technology,Zhengzhou 450007,China;School of Computer Science,Zhongyuan University of Technology,Zhengzhou 450007,China;School of Cyber Science and Engineering,Zhengzhou University,Zhengzhou 450001,China)

机构地区：[1]中原工学院软件学院,河南郑州450007 [2]中原工学院计算机学院,河南郑州450007 [3]郑州大学网络空间安全学院,河南郑州450001

出　　处：《中原工学院学报》2025年第1期1-11,共11页Journal of Zhongyuan University of Technology

基　　金：河南省科技攻关项目(242102210060)。

摘　　要：针对传统随机森林模型在软件缺陷预测领域中预测精度低和参数优化困难等问题,提出了一种基于改进蛇优化算法(ISO)优化随机森林模型的软件缺陷预测方法。蛇优化算法改进过程:首先采用莱维飞行策略初始化种群位置,增强种群的初始多样性;其次引入基于正余弦扰动因子的更新策略和基于自适应机制的幼蛇变异策略,增强局部搜索能力并避免陷入局部最优;最后通过在11个基准测试函数上与原始蛇优化算法及4种经典算法比较,以评估ISO算法的寻优性能。结果表明ISO算法具有较高的收敛精度和更强的稳定性。而面对软件缺陷预测领域中存在的类不平衡问题,进一步地使用ISO算法优化随机森林模型参数,并采用ISO-RF算法提高随机森林模型的软件缺陷预测精度。通过在3个项目的10个公开数据集上的对比实验结果表明,ISO-RF算法在召回率、F1值、马修斯系数MCC等评价指标上明显优于其他对比算法。To address the challenges of low prediction accuracy and difficulty in parameter optimization faced by traditional random forest in the field of software defect prediction,a refined method called ISO-RF is proposed,utilizing an improved snake optimization algorithm to optimize random forest model.The improvements to the snake optimization algorithm are as follows.Firstly,the population positions are initialized using the Lévy flight strategy to enhance the initial diversity of the population.Next,an update strategy based on the sine-cosine disturbance factor and a mutation strategy for the young snakes based on an adaptive mechanism are introduced to enhance the local search capability and avoid falling into local optima.The performance of the ISO in optimization is evaluated by comparing it with the original snake optimization algorithm and four classic algorithms across 11 benchmark test functions.The results indicate that the ISO algorithm achieves higher convergence accuracy and greater stability.Furthermore,addressing the class imbalance problem prevalent in the field of software defect prediction,the ISO is used to optimize the parameters of the random forest model.The proposed ISO-RF method enhances the prediction accuracy of the random forest model.Experimental results on 10 public datasets from three projects show that the ISO-RF algorithm significantly outperforms other comparative algorithms in terms of recall,F 1-measure,and MCC,demonstrating certain promotional value.

关 键 词：软件缺陷预测 改进蛇优化算法 随机森林 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]