Font Size: a A A

Based On Rough Set _ Software Risk Evaluation Model Of Neural Network Research

Posted on:2013-08-02Degree:MasterType:Thesis
Country:ChinaCandidate:X Y LiuFull Text:PDF
GTID:2249330377456307Subject:Industrial Economics
Abstract/Summary:PDF Full Text Request
With the rapidly advance and update of information technology, the outsourcedsoftware industry has also been an unprecedented development, the demand ofsoftware is increasing. The reason is that many companies want to improve businesscompetitiveness and economic efficiency. Software project has been a high riskinvestment projects combined with the high complexity of outsourced product, so therisk of software project can ‘t be ignored. Accurate identification, assessment andtreatment of risk in the processing of outsourced software development are theeffective methods to improve the success rate of IT project. Throughout the softwarelife cycle, risk factors are very complex and diverse, and have the characteristics ofincomplete and inaccurate, and the amount of data is also continued growing. Inthis case, by studying the relevant literature, I have proposed a software project riskassessment model which based on the theory of combination of rough sets and BPneural network. The model aims to help outsourced software company make the rightdecisions on the beginning stage of project. The model is a combination of weakcoupling of rough sets and BP neural network. Rough set is a major role in the datadimensionality reduction by reducing attribute and reducing value to the sample set.Rough set can simplify the network topology of the BP_neural network structure,which can shorten the training time of BP neural network and can improve theaccuracy of the model predictions. This paper first reviews the domestic andinternational IT project risk management status and related models, and analysis andcompare several traditional risk identification methods. Then through researchingliterature, visiting to several outsourcing software companies to build IT project riskfactor indicators and historical sample data. Finally, training the model, and verifythat the model is superior to single BP neural network model.
Keywords/Search Tags:IT project risk, BP neural network, rough sets, risk assessment
PDF Full Text Request
Related items