Defect Prediction based on Machine Learning using System Test Parameters

Since its inception, software testing has evolved over the years. In this era of Artificial Intelligence & Machine learning, there are many recent advancements in Software Engineering. Various software engineering metrics are analyzed; deciphered and the required predictions are made. Defect prediction is one such activity which is of great significance in improving the software quality which helps software developers and testers to focus on the modules which are more likely to defect prone. In the literature survey, we found several techniques to predict defects based on features which were mostly captured from the development perspective. In this paper, we propose our machine learning based approach with the main objective of finding potential areas of the defects by considering parameters from System Testing matrices & a unique `parameter called Component Dependency score'. This method also helps in controlling the software quality of a dynamically evolving software.