Incorporating unsupervised machine learning technique on genetic algorithm for test case optimization

Abstract EN

Search-based software testing uses random or directed search techniques to address problems.

This paper discusses on test case selection and prioritization by combining genetic and clustering algorithms.

Test cases have been generated using genetic algorithm and the prioritization is performed using group-wise clustering algorithm by assigning priorities to the generated test cases thereby reducing the size of a test suite.

Test case selection is performed to select a suitable test case in order to their importance with respect to test goals.

The objectives considered for criteria-based optimization are to optimize test suite with better condition coverage and to improve the fault detection capability and to minimize the execution time.

Experimental results show that significant improvement when compared to the existing clustering technique in terms of condition coverage up to 93%, improved fault detection capability achieved upto 85.7% with minimal execution time of 4100ms.