Regression Testing Prioritization Technique Based on Historical Execution Information

Abstract

Regression testing is a very important and effective testing method in the software life cycle, and its cost accounts for about 70% of the cost of software testing. Typically, regression testing re-runs all existing test cases in a random order, but this is sometimes not feasible because it leads to high test resource consumption and long feedback cycles. In this paper, we focus on how to obtain and prioritize test cases with a high fault detection rate based on historical execution information. Secondly, how to improve the efficiency of priority technology based on historical data in continuous integration. We implemented a series of experiments using three industrial datasets. The experimental results show that the prioritisation techniques based on average historical failure rates and average historical execution failure rates outperform other techniques; If historical execution information is insufficient, techniques that integrate historical execution information with requirements priorities provide better performance.