ML Based SMART Test Script Optimizer

In Growing Business Scenarios, product portfolio keeps increasing and so also the complexity of products. This brings us to a scenario where product stability validation becomes a challenge, as we need to optimize the ways of qualification with new feature & portfolio additions. In order to keep the qualification pipeline efficient, we present here a novel approach which utilizes machine learning based techniques to optimize test script selection towards effective test planning. This method increases the efficiency in terms of time and effort needed for qualification. Test script optimization involves focusing on providing solutions to avoid duplicate effort involved due to unintended redundant test scripts being part of the execution and an efficient planning which can help improving API coverage rate without impacting the existing API coverage. This paper proposes two Machine Learning based solutions for test script optimization. First one is Redundant Test Script identification solution based on ElasticSearch More Like This query, which helps in duplicate test script elimination using lucene scoring formula. Second one is Test Script Clustering Solution based on unsupervised machine learning (Bisecting K-Means algorithm) to form cluster of test scripts depending on similarity among documents. Which helps in making efficient plan with improved API coverage rate by touching unique API calls by left shifting with quick turnaround time of execution cycle. In turn it helps in early defect discovery and improves product delivery timeline.