外包软件项目风险决策树智能分析模型

Intelligent Decision Tree Model for Software Outsourcing Project Risk Analysis

软件项目外包已成为主要的软件开发方式,但风险却很高.本文提出了外包软件项目风险的决策树智能分析实证模型,并用此模型对项目风险进行评估和控制.建立了基于客户方和承包方双视角的风险识别概念模型,收集了外包软件项目真实样本用于决策树风险智能分析模型的训练和验证.实验结果表明,本文所采用的方法在准确率方面优于神经网络、朴素贝叶斯算法.决策树模型所发现的管理规则与软件工程理论相吻合,能很好地指导项目风险评估分析.

Software outsourcing is the major means of software development,which is of higher development risk.An intelligent Decision Tree risk analysis model for software outsourcing projects is useful for risk assessment and control.First,we established a risk identification theoretic model based on distinguished views of customer and contactor.Second,we collected real software outsourcing project samples to train and verify the Decision Tree model.Experiment results indicate that the proposed Decision Tree model outperforms Neural Networks and Nave Bayes in terms of prediction accuracy.Management rules derived from this analysis model are in conformity with software engineering theories,thus they can be regarded as admirable guidance of risk assessment.This model is the first empirical model based on real software outsourcing project samples in China,which should be a great guideline of software outsourcing project risk management of China.