Association Rule Mining And Its Application In Software Project Risk Management

Improving the level of information comprehensively is clearly to put as one of the SixInitiatives to develop the modern industrial system and improve the core competitivenessof industries by The National Twelfth Five Year Plan. We will develop and enhance thesoftware industry, and strengthen the construction of critical information systems.However, the software projects have brought great risks to the enterprises developing andneeding them with the constantly updating software development technology, the growingnumber and the increasing complexity. With the arrival of the Big Data Era, thecontradiction of the Rich Data and Poor Knowledge more urgently appeal to a wide rangeof applications of Data Mining Techniques. The subject start with this need and used theAssociation Rule Mining technology and the principle of Bayesian Networkcomprehensively to research in the field of the Software Project Risk Management. Theobtained association rules were used to guide software project risk management practices.Based on this, we conducted the innovation on the data sources of the Association RuleMining and created a new idea of software project risk management.We built the software project risk management process model and introduced theBayesian Network theory combining the existing identification methods of risk factors.Based on the Bayesian Network, we established network model of risk factors. Accordingto the principle of the Bayesian Network, we reasoned and calculated on the priorprobabilities combining qualitative and quantitative analysis. After rigorously rationalcalculation and analysis, we obtained a large number of risk data providing accurate andreasonable data source for the implement of Association Rules Mining. It has broken theprevious mining methods based on the direct historical data that using of the BayesianNetworks and the Association Rule Mining integrated. In addition, it also innovated a newidea of getting the data source for Data Mining and made up for the shortage of historicalrisk data and the defects mainly relying on qualitative analysis that preprocessing on therisk data obtained by the Bayesian Network. Based on the platform of SQL Server2005,we provided a guarantee for the effectiveness of the software project risk managementmore accurately by using the Analysis Services (SSAS) to realize Association RulesMining. It is not only a innovation for the application of the Association Rule MiningTechnology, but also a pioneering for the software project risk management techniquesand methods that applying the Association Rule Mining to the field of Software Project Risk Management. We can discover association rules by mining on the risk data and makesemantic analysis on this risk rules so that the semantic of risk rules can be completelydescribed. Based on this, we put forward a systematic framework of the rules descriptionand rationalization proposals on Software Project Risk Management.