The Application Of Defect Prediction In Test Case Prioritization

The rapid development of computer technology drives the increase of smartphone users,all kinds of software have flooded people's lives.Different needs of different people promote software development.When the software meets the basic requirements,everyone starts to pay attention to software quality.Software testing technology ensures the quality of software products,which is mainly attributed to the application of regression testing technology.But in the traditional regression test,it faces a lot of resource consumption,causing excessive waste of manpower and time.The test case set is maintained through the test case prioritization technology,and the defect prediction technology is introduced into the test case prioritization technology.By enhancing the relationship between the sorting target and the sorting result,the purpose of increasing the early defect detection rate and saving development costs is achieved.A Aiming at the slow convergence problem of genetic algorithm in single-objective sorting,and the one-sided evaluation of test cases caused by the sorting results generated under single-objective.First,the test cases are reduced,and secondly,genetic algorithm is applied to multi-objective sorting,and the two targets jointly guide the generation of sorting results.Experiments show that the optimized test case sorting method has a significant improvement in the values of evaluation targets such as the average defect detection rate.There is no strong correlation between currently commonly used ranking targets and defects.Introducing defect prediction technology into the test case ranking method,Propose a test case ranking method based on defect prediction and Choose average defect coverage as the sorting target.Strengthen the relationship between the sorting target and the defect,and improve the defect detection rate.While applying defect prediction technology,address the problems of the data set in the defect prediction technology.First,use the LLE method to reduce the dimensionality of the data through feature selection.and then use the method of combining the SMOTE and the undersampling technique to deal with the class imbalance problem.The experiment evaluates the model through evaluation indexes such as precision rate,and comparative experiments show that the improved method has obvious advantages in evaluation indicators.The existing problems and introduced problems are focused on in the prioritization of test cases from the perspective of the sorting goal and the sorting method.The comparison of experimental results shows that the proposed method can better guide the sorting of test cases to a certain extent,and optimize the generation of sorting results.