Performance improvement model of regression test selection

Abstract

The size of test case reduction using test cases of the improved program is considered as a primary purpose of developing various regression test selections. This also concerns with fault avoidance that could be found from some test cases selections. This study focuses and compares the proficiency of solving these two problems using the three methods from previous studies; original regression test, code coverage based, and filtering-based selections; and a new proposed model. This model carries on five processes; 1) testing test suit and looking for passed or failed test cases, 2) correcting failures that can be restored, 3) categorizing all passed test cases into four areas: user, system, functional, and non-functional cases, 4) taking out irrelevant objects, 5) selecting the proper test case. The study found that the result of the size reduction using the proposed model is much larger than the existing methods by 2.49%-8.55%, while the percentage of avoiding faults using the existing algorithms are lower than the new proposed algorithm around 0.58%-2.36%.