GitHub Classroom commit records before and after generative AI: replication package (activity attribution, platform change, organization weighting)
收藏资源简介:
Replication package for a study of commit records in public GitHub Classroom repositories (58,516 repositories collected in August 2026, shortly before GitHub retired Classroom). It contains the collection, extraction, and analysis scripts; hashed derived tables at the commit and repository level; the result file of every analysis; the LaTeX source of the manuscript; an entry point that rebuilds the generated tables, figures, and text numbers from the result files, re-fits from the derived tables the models of the main comparisons and of their variants, and checks the numbers typed into the manuscript; the preregistered protocol and its results; the post-registration decision log (D1 to D10, E1 to E15); and a log of generative-AI use. Raw commit logs and a manual-review file containing commit text are not distributed.



