Small And Medium Software Enterprises In The Project Risk Mechanism Analysis And Early Warning

Because of the demand caused by implementing the competition strategy of country inside and outside, the development of the middle and little enterprises' projects in our country arrived the historical climax in the scale, genre and quantities. But the software projects' risks didn't be softened with the improvement of the technology, the mature of the application. Instead, the expansile trend is extremely obvious. So it is very essential for our young middle and little enterprises to collect the information, to analyse the projects' risk factors and the action mechanism between the factors influencing the project performance, to strengthen research towards the risk beforehand alarm, to intensify the enterprises' risk consciousness.23 key risk factors in projects of the middle and little software enterprises are found out by empirical study, which reveals the uniqueness of risks in projects of the middle and little software enterprises. Many conclusions of study on risk factors in common IT software projects cannot be applied in projects of the middle and little software enterprises directly.The path relationships from risk factors to risk consequences in projects of the middle and little software enterprises, which describe the risk generation mechanism in projects of the middle and little software enterprises, are explored with Structural Equation Modeling (SEM). The results of data analysis reveal that not all risk factors have direct effects on risk consequences, and the changes of several risk factors influence the consequences via some other factors indirectly.Based on the analysis results mentioned above and the basis of understanding the producing mechanism of the risks, a risk alarm model that can predict the degree of risks seriousness in projects of the middle and little software enterprise is constructed with the ANN. We use the computer as the tool, write a BP NN program with the neural network toolbox in the MATLAB and make a simulator to calculate the risks in middle and little software enterprises' projects. By using this simulator in ten projects, we found the result is good, which means this methord is effective. So we offer a new research method for the beforehand alarm towards the software projects' risks in this paper.