Abstract: In order to enhance the estimation precision and reduce the difficulty of project selection by manpower, a software cost estimation method based on genetic algorithm and case-based reasoning is proposed. The estimation steps of case-based reasoning are given and the similarity function is constructed. The genetic algorithm for case-based reasoning is designed. According to the algorithm, the similar cases are selected in the history dataset, and the feature weighting of software project is optimized. The proposed method is analyzed by the Albrecht and Desharnais dataset. The experimental results show that the new method can enhance the estimation precision of software cost in the early of software lifecycle. Comparing to the analogy method and linear regression method, the new method has less mean magnitude of relative error. The estimation performance can meet the software estimation requirement.

Key words: software cost estimation, genetic algorithm, case-based reasoning, feature weighting

摘要： 为了提高类比法的估算精度，减少人工检索案例的工作量和难度，提出了一种基于遗传算法和案例推理的软件费用估算方法。给出了案例推理过程的估算步骤，构造了案例的相似性度量函数；设计了用于案例推理问题的遗传算法，利用该算法在历史数据库中搜索与目标案例最相似的项目，并对软件项目的特征权重进行优化；借助Albrecht和Desharnais数据库，对提出的方法进行分析。实验结果表明，该方法可以在软件生命周期的早期显著提高软件费用的估算精度，与类比估算和线性回归方法相比，具有更小的平均误差率，其估算性能符合相关行业软件的估算需求。

关键词: 软件费用估算, 遗传算法, 案例推理, 特征权重