The Research Of IT Project Risk Assessment Model Based On Neural Network

With the development of science and technology, enterprises pay more attention to informatization construction, in order to complete business process reengineering and improve work efficiency. Under this background, a large number of IT development projects started in recent years. Meanwhile, the success rates of IT development projects were very low. After spending a lot of money, time and human resources, the software and information system developed cannot meet the needs of users, and brought tremendous loss to both users and enterprises.The IT project has a long cycle, large investment, rapid technological update and is full of uncertainty, which causes high risk of IT project to some extent. But in reality, to speed up the progress,enterprises usually ignores the risk assessment. A lot of risk which should had been avoided occurred, and eventually lead to the failure of project. Therefore, risk assessment plays a important role in the area of IT development projects.Aimed at the problem of high risk for IT projects, this paper proposes a model which combines Artificial Neural Network and Principal Component Analysis theories, to assess the risk for IT projects and provide decisions for managers. This paper mainly focuses on the following study and research:Firstly, using the theories of project risk management combining with the process of IT project development, to recognize the risk factors of IT project and establish the index system of IT development project.Secondly, on the basis of IT risk assessment index system, establishing risk assessment model based on Artificial Neural Network and Principal Component Analysis. The model firstly normalizes the sample data and simplifies the neural network training. And then using the Principal Component Analysis, to reduce the dimension of input samples and improve the efficiency of neural network training. Taking the index after Principal Component Analysis as the input of neural network, determining the parameters of the network.And then using Matlab neural network toolbox for simulation, analysis the results according to the angle of error and training rate, verifying the feasibility of the model.Finally, designing the application based on the simulation.Firstly, proposing the whole structure of system, and analyzing the business processes, then desiging the dababase and functions.Finally, realizating the function of system.