The Research Of Software Project Risk Assessment Based On BP Neural Network

In the present fields of the social life, the software becomes more and more vital and atthe same time software industry has become a kind of the energies accelerating the growth ofthe national economy. As a large systematized project, the software project has thecharacteristics of high complexity, high technology and frequent renovation, which thendetermine high risks of it. Additionally, with the enlargement of the modern software projects,the control and management appear extremely difficult and the projects are often confrontedwith the problems of overspending and slow progress. Consequently, the success rate of theexploitation of them becomes extremely low. In order to cope with the disadvantageoussituation of the exploitation, reduce uncertainty of the projects and the consequent loss itproduces and ensure the realization of the project objective, risk management has beenintroduced into the domain of software. In the theory of risk management, risk assessmentoccupies a very important position, which is the basis of risk control. So, when carrying outthe risk management of the software project, we must locate our great attention on the riskassessment, which is just the object talked about in the present thesis.The thesis, first of all, summarizes the present research situation of risk management ofsoftware project at home and abroad and introduces its basic theory; next, on the basis ofexpatiating on the methods of risk recognition, combining software project's WBS and thenine knowledge areas of project management, the present thesis has identified the basic riskfactors, and built the risk assessment index system of the software; and then, by contrastivelyanalyzing the current leading methods of risk assessment, this thesis chooses artificial neuralnetworks as its assessment method, and constructs a software project risk assessment modelbased on the BP neural network; finally, the thesis makes a training test and an appliedanalysis of the specific sampled data to testify to the built assessment model.