Research And Application Of Test Case Prioritization Based On Regression Testing

Software testing is an important aspect of software development and a very important means to assure the quality of software. Regression testing, which is performed after changes of a software system or before release of new versions of the system, is a commonly used method of software testing. However, testers usually have no enough time and resources to test software because of release of new version early. Therefore, improving the efficiency and effectiveness of regression testing to reduce the cost is the focus in the field of research of regression testing.Tests need to manage test suites uniformly to increase the efficiency of regression testing. Schedule test cases in order to meet some performance goals, such as code coverage, rate of fault detection, etc.Test case prioritization can improve the performance of test suits and do not discard test cases. Currently, many researchers of the world have made some valuable achievements on test case prioritization, but many techniques are to be further researched.Existing test case prioritization techniques are lack of dynamic.On one side, the testing requirements not yet covered are changing frequently during regression testing and the test case priorities are also changing. On another side, existing techniques do not consider testing requirements, test case costs and benefits.As a result, we present a new cost-benefit model based on varying testing requirements and an algorithm-TCP_Req_CB_add. In order to demonstrate the effectiveness of the algorithm, we also present an associated metric.The algorithm indicate how to calculate testing requirements priorities, test case costs and benefits based on testing history information,adjust test case priorities according to varying testing requirements and the matrix of testing requirements and test cases, and run test cases. It considers factors fully, so it's reasonable.At the last of the paper, we use the method in Power Data Analysis System and evaluate the result of test cases order using the new metric.Compared with other algorithms, the new algorithm is more efficient.