Combinatorial-Based Event Sequences for Reduction of Android Test Suites

Abstract

Several studies show that combinational-based reduction techniques that consider user event combinations and sequences have the potential to reduce regression testing costs while maintaining high code coverage and fault finding effectiveness for desktop GUI and web-based applications. This work expands existing techniques to the mobile testing domain by utilizing not only user event coverage, but additionally element coverage. We use sequences of size $t=2$ to reduce test suites by user event coverage and element coverage using three scenarios for each, including sequences without respect to order of occurrence, sequences with respect to order of occurrence, and sequences of consecutive occurrences only. The results demonstrate that reductions guided by event sequences reduce the test suite by between 24.67%-66% while losing at most 0.39% code coverage. Element sequence guided reductions reduce the test suites more dramatically by 40% to 72.67%, losing less than 0.87% code coverage.