Software Testing Defect Data Analysis And Management

Software defect analysis in software development and software testing plays a very important role, not only can reduce the development cost, but also improve work efficiency, so it is necessary to defect analysis.This article first gives a detailed overview of the software defects, but also introduced a software defect analysis of current situation and the need to address the problems and difficulties. In the theory of knowledge inside the rough set theory and Bayesian methods of network theory, and a detailed description of the two methods, respectively, to analyze how software defects.Followed the ideas of the analysis and management of software defects on rough sets and Bayesian network analysis combined. Specifically, the first rough set of defect data can be simplified; the benefits of such a treatment can be avoided analysis of the excess property and to simplify the calculation steps. Defect data after processing in accordance with the defects in the source is divided into the following four levels:requirements analysis phase, design phase, coding phase, and software testing stage. Then combined with the Bayesian network analysis to the establishment of a Bayesian network model for each stage, and ultimately the use of Bayesian networks to analyze the processing defect data. Finally, we analyze the benefits from the combination of these two methods use the instance of the software design phase in the project development system, in accordance with the idea of our defect analysis using a combination of rough reduction and Bayesian network modeling, each node in a given probability distribution, we began to analyze evidence of defect has occurred in front of each influencing factor defects in the evidence of the size of the data analysis we can conclude that the defects affect the probability of each node, the data analysis we can for the subsequent software development and testing of proposed changes. In order to calculate conveniently, through Matlab FullBNT-1.0.4toolbox programming to create the defect model of the design phase, and use the model to deal with the marginal distribution and joint distribution, the conditional probability distribution of the simulation, the results to analyze.In this paper, the Bayesian network on the defect data analysis and processing, defect analysis can not only play a guiding role of software testing, effective relationships and analysis, but also help improve the development and testing process, and enhance product stability, earlier found that the risk module, and ultimately improve efficiency and reduce development costs.