Funded by TUBITAK 1505 (‘’Yazılım projelerinde geliştiriciye özel hata tahmini ve hata çözüm süresi tahmini modellemesi”, April 2018 – Sept 2019), the project aims to build AI-based personalized models for software developers that would guide them through finding bugs in their code modifications, and predicting bug resolution times. In collaboration with Ericsson Turkey, version control systems, test design and execution tools, and issue repositories of the selected projects were crawled to identify features that likely predict bugs in code modifications. General bug prediction models, personalized models and their combined versions have been designed to empirically assess the effect of personal development characteristics on bug injection rates. Another AI-based model has been implemented to predict resolution time of bugs reported in the issue repositories using bug attributes and textual descriptions.
