摘要
为了解决软件风险分析中可能出现的数据不完整以及影响因素间关系复杂的问题,提出了一种改进贝叶斯网络的软件项目风险分析方法。将遗传算法和EM算法相结合得到EM-GA算法,利用EM-GA算法对软件项目分析过程中贝叶斯网络结构中的参数进行学习,同时优化网络结构,通过实例验证了该方法的有效性及可行性。
In order to solve the problem of incomplete data and complex relations among influencing factors which may appear in the software risk analysis,this paper presented a software project risk analysis process based on Bayesian networks which has been improved.Firstly,presented a EM-GA algorithm based on genetic algorithm.Then,used the algorithm to optimize the Bayesian networks structures and solve Bayesian parameter learning.Finally,the experiment results show this algorithm provide a new method for software project risk analysis process.
出处
《计算机应用研究》
CSCD
北大核心
2010年第4期1360-1362,共3页
Application Research of Computers
基金
国家教育部"新世纪优秀人才支持计划"资助项目(NCET-07-0908)
