Software outsourcing is one of the leading methods in software development. However, it is also accompanied with higher risk than in-house software development. A risk intelligent analysis model based on Bayesian Network can effectively contribute to software project risk assessment. From the perspectives of both the customer and contractor, we propose a risk identification framework for outsourced software projects, and have collected real-life outsourced software project samples. Based on totally 154 valid samples, we established an intelligent analysis model for outsourced software project risk by incorporating expert knowledge as structural constraints into a Bayesian Network. Experimental results showed that the model has higher predictive accuracy than Decision Tree and Neural Network, and the derived management rules are consistent with the existing software engineering theory. The model would provide a great guideline for outsourced software project risk management in both theory and practice

No abstract