It Outsourcing Risk Assessment Model Based On PSO_GA Fuzzy Neural Network

The rapid development of IT outsourcing in the past decade brings many benefits, such as reduced IT costs, increase management efficiency,focusing resources on core competencies, etc. However, IT outsourcing also has a lot of risk.IT outsourcing risk assessment is an important part of enterprise IT outsourcing risk management, risk assessment method is correct or not, will directly affect the risk management. Having methods are too dependent on the experts, the evaluation results are affected expert subjective preferences, knowledge structure, determine the level, so the evaluation is not enough scientific. Based on previous research results of the risk assessment method, risk evaluation method is proposed based on factor analysis and fuzzy neural network,to establish enterprise IT outsourcing risk assessment model, so that enterprises can use this model to assess risk.Firstly, on the principle of scientific, systematic, feasibility, importance, establishing enterprise IT outsourcing risk evaluation index system through reading, summarize the literature, and then,with so many indexs,we do factor analysis based on questionnairethe, to get a comprehensive evaluation of risk factor reduction evaluation to provide a basis for the establishment of evaluation model.Secondly,using fuzzy neural network with the function of non-linear mapping and handling fuzzy data,combined with the integrated factor to construct the enterprise IT outsourcing risk assessment model. This model, although it was found to some extent in improving the objectivity and accuracy of the evaluation, but the evaluation of the efficiency and accuracy need to be improved, so we choose studying fuzzy neural network learning algorithm and sample selection, propose a PSO_GA learning algorithm and sample selection method based on the weighted distanceto to make the evaluation model more efficiently and exactly.Finally, studying the project of the website integration outsourcing to prove the feasibility of the model.