Prioritizing Test Cases with Markov Chains: A Preliminary Investigation

Test Case Prioritization reduces the cost of software testing by executing earlier the subset of test cases showing higher priorities. The methodology consists of ranking test cases so that, in case of a limited budget, only the top-ranked tests are exercised. One possible direction for prioritizing test cases relies on considering the usage frequency of a software sub-system. To this end, a promising direction is to identify the likelihood of events occurring in software systems, and this can be achieved by adopting Markov chains. This paper presents a novel approach that analyzes the system scenarios modeled as a Markov chain and ranks the generated test sequences to prioritize test cases. To assess the proposed approach, we developed an algorithm and conducted a preliminary and experimental study that investigates the feasibility of using Markov chains as an appropriate means to prioritize test cases. We demonstrate the strength of the novel strategy by evaluating two heuristics, namely H1 (based on the transition probabilities) and H2 (based on the steady-state probabilities), with established metrics. Results show (i) coverage of 100% for both H1 and H2, and (ii) efficiency equal to 98.4% for H1 and 99.4% for H2, on average.