基于最小二乘支持向量机的军用软件开发成本测算研究

Research on Prediction of Military Software Development Cost Based on LS-SVM

利用最小二乘支持向量机(least squares support vector machines,LS-SVM)适用于小样本预测的优势,将该方法运用到军用软件开发成本测算领域,建立了基于LS-SVM的军用软件开发成本测算模型并进行了实例分析。结果表明:在同样的样本数据条件下,该模型有效提高了测算精度,为军用软件早期开发的成本测算提供了新的有效方法。

Using least squares support vector machines （LS-SVM）,the method is applied to mil-itary software development cost estimation because of its advantages in small sample prediction.The LS-SVM model of military software development cost prediction is established,and the model with the actual data is tested.The results show that under the same sample data condition,the model in-creases the accuracy exactly,and provides a new and effective method for military software develop-ment cost prediction.