Abstract: Software defect prediction is to predict whether there is a bug in a software system module. Traditional researches on software defect prediction mainly focus on supervised method. This type of method needs a lot of instances with labels as the training set. However, in engineering practice, the instances with labels are difficult to obtain. This paper proposes an improved fuzzy C-means algorithm, combining simulated annealing and genetic algorithm, to solve the defect that fuzzy C-means is easily affected by the initial cluster centers. The experimental results show that the proposed method has high robustness and accuracy on software defect prediction.

Key words: software defect prediction, fuzzy C-means, simulated annealing, genetic algorithm, unsupervised

摘要： 软件缺陷预测用来预测软件系统各个模块中是否存在BUG。传统的软件缺陷预测技术研究主要局限在有监督方法上，这类方法需要大量的已标注数据进行训练，但在工程实际中，这类标签数据不易获取。提出了一种结合模拟退火和遗传算法的改进模糊C均值算法，以解决模糊C均值容易受初始聚类中心影响而收敛到局部最优的缺陷。实验结果表明提出的方法在软件缺陷预测中具备高鲁棒性和较高预测精度。

关键词: 软件缺陷预测, 模糊C均值, 模拟退火, 遗传算法, 无监督