摘要
提出一种改进的软件项目投资风险评价模型。该模型在软件项目投资风险评价指标体系的基础上采用因子分析的方法进行降维处理,从而减少问题分析的维度。同时,在B-P神经网络的建模过程中利用一种基于黄金分割原理的优化算法确定隐含层节点数,提高了风险评价的精度。实际评价数据表明,该模型能够有效地完成软件项目投资风险的评价。
An improved risk evaluation model in software project investment is presented. Based on the evaluation index system, the model utilizes factor analysis way to decrease dimension of sample data for simplifying problem. In the modeling of Neural Network (NN), an optimization algorithm based on the principle of golden section to design the number of hidden layer nodes is utilized for increasing the precision of risk evaluation. The model can accomplish the work of risk evaluation in software project investment by verifying the sample data of risk evaluation.
出处
《计算机工程》
CAS
CSCD
北大核心
2008年第7期50-52,55,共4页
Computer Engineering
基金
国家自然科学基金资助项目(70371046)
关键词
软件项目
风险评价模型
因子分析
神经网络
隐含层
software project
risk evaluation model
factor analysis
Neural Network(NN)
hidden layer