Genetic Algorithm Based Statistical Regression Test Data Generation Method

Abstract

The invention discloses a statistical regression test data generation method based on genetic algorithm, which aims at rapidly and high-efficiently generating the test data for covering a target path. The method comprises the following steps of (1) determining relevant input variable according to an alter statement in the regression test process; (2) modeling the probability distribution of the input variable associated to the alter statement; (3) forming initial population on the basis of partial probability distribution of the an input component in the program before the alter and a distributed value; (4) calculating an applicability value of each individual in the population according to a designed applicability function; (5) judging whether an optimal solution is found for a target function according to the calculated individual applicability, and outputting the test data and entering the step 6 if the optimal solution is found; otherwise, carrying out variation operation for the individual to generate a novel individual and returning to the step 4; and (6) combining the probability distribution of the input variable which is not related to the alter statement to obtain the probability distribution of all input variables; and generating the test data according to the distribution sampling.

Description

technical field [0001] This patent belongs to the field of software testing, and specifically relates to a method for generating statistical regression test data based on a genetic algorithm, which can be used in software testing to generate test data covering target path sets. Background technique [0002] Software testing is the key to software quality assurance, an indispensable link in software development, and an important part of software engineering. The effect of testing is directly related to the quality of software products. According to the statistics of the American industry, software testing accounts for more than 50% of the total cost of the entire software development cycle, including many links that are very time-consuming, boring, very difficult and error-prone. In addition, with the continuous improvement of the complexity of computer control objects and the continuous enhancement of software functions, the scale of software is also continuously increasing....

AI Technical Summary

Problems solved by technology

[0005] The difference from traditional testing is that statistical testing can estimate the reliability of the program. In other testing methods, the test cases are carefully designed and screened, and the reliability of the program cannot be evaluated.

Images