refactoring-dataset

This dataset contains 162,737 verified refactoring instances mined from 60 open-source Java projects across three distinct ecosystems: Apache Software Foundation (20 projects), GitHub (20 projects), and F-Droid (20 projects). The refactorings were detected using RefactoringMiner 3.0.10, which achieves approximately 99% precision and 94% recall. Key Features Verified instances: All refactorings detected by RefactoringMiner 3.0.10 with high precision Rich context: Source code snapshots from parent commit (before) Balanced across ecosystems: 20 projects from each of three distinct open-source communities Long-tail distribution: Realistic imbalance with head classes dominating (62-85%) and tail classes representing <5%