基于灰色关联分析的软件工作量估算方法

针对历史数据少、信息贫乏的软件估算问题，提出了一种基于灰色关联分析的工作量估算方法。首先结合灰色关联理论分析了项目特征与工作量的关联度，然后利用回归技术选取最优特征集；在此基础上，计算新项目和历史项目的关联度，并根据关联度大的项目估算软件工作量；最后通过4个典型的软件数据集对估算方法的性能进行分析。实验结果表明，该方法能够在历史数据少的情况下准确估算出软件工作量，其性能优于基于回归分析、BP神经网络和类比估算等方法；而选取最优特征后的灰色关联方法，由于剔除了与工作量相关程度低的特征因子，进一步减小了工作量估算的平均误差率。

Software effort estimation method based on grey relational analysis

Under some small history data condition, a grey relational analysis method for software effort estimation is proposed. Firstly, the relationship grade between features and effort is analyzed according to the grey relational analysis theory. At the same time, the optimal feature set is chose by regression technique. On the basis of which, the grey relationship grades between new software and history dataset are computed, and some high relationship grade software data are used for estimating the effort of new software. Finally, four history datasets are used for analyzing the performance of the estimation method. The results of experiment show that the proposed method could precisely estimate software effort, and the method is superior to the linear regression, back propagation network, and analogy method. After some low relationship grade features are deleted from the optimal feature set, the mean magnitude of relative errors is further reduced.