Test Case Evaluation Model

With the popularity of computers and applications, the human is increasingly dependent on software. And at the same time, the issue of software quality has brought us more and more serious harm, making it more and more attention to the quality of the software.Software testing is an important software quality assurance techniques, which occupied a very prominent position in the software life cycle. According to Boehm's statistics, the cost of testing will be accounted for 30% to 50% of the total cost of software development. Therefore, enhancing the efficiency of software testing and reduce test cost are of great practical significance.Test cases are the core of the whole software testing. A good test case will not only find errors as many as possible, but also find the errors have never be found. Then the efficiency of testing is improved. The test data is the most important part of the the test case, so the key of a good test case design is to design a good test data, which plays a crucial role in the whole process of software testing. The synthetical evaluation for the test cases is helpful to select the good test case and reduce the testing cost.This dissertation first introduces the study of the test cases evaluation at home and abroad. At present, the coverage is the main criteria for sort and selects test cases, including the code coverage, branch coverage, condition coverage. And the errors and consumed time during the testing are also used to sort and select test cases. However, it did not consider these factors to form an overall evaluation model. Aim at these problems, this dissertation presents five factors that affect the test case, including the reuse times of the test data, the ratio of testing time to interface size, the coverage of test cases, the severity of the error found by test cases and whether find the errors have never be found. At first, the Analytical Hierarchy Process method is used to calculate the weight value of the above five factors, and the initial synthetical evaluation model is proposed. Then, the value of the five factors under different conditions are determined by using the method of Multiple Comparisons and Dynamic Weight Through the experiments, the similarity of the test case sort order between the result calculated by the model and the experts is determined to evaluate the model. The weight value is determined by the similarity. The Optimal Solution can be determined by the minimum of Variation, which is the weight value of the factors. At last, the synthetical evaluation model for the test cases is proposed.