Tosin D. Oyetoyan, Daniela S. Cruzes, and Reidar Conradi
Component, Smart Grid , Prediction Model, Empirical Study, Component Defect-Proneness
Smart Grid systems are kind of System of Systems with distributed and highly heterogeneous software connected to provide various services. Early knowledge of defect prone parts is useful for improving the safety and maintenance of these kinds of systems. We report the results of an empirical study of using reused components to predict future defective components in a type of open source Smart Grid application (transmission and operation domain of Smart Grid). Our results showed that reused components of this Smart Grid application are strong predictors of future defective components. The model’s best predictors gave an average recall of 0.92 (average precision of 0.406) when tested across three future releases. Which implied that 92% of predicted defective components in the next release turned out to be defective. This model can be employed to tailor quality assurance (QA) efforts in a way that blind spots are avoided in such critical system and QA effectiveness significantly improved
Important Links:
Go Back
