GitHub Pull Request Quality Dataset

A dataset of 79,058 merged pull requests from 24 popular open-source GitHub repositories with 142 engineered features for predicting PR review time. Features include code complexity metrics, NLP sentiment analysis (VADER, TextBlob), readability scores (Flesch-Kincaid, Gunning Fog), author experience profiles, and PR quality indicators. Target variables: review_time_hours (regression) review_speed_category (classification: fast/medium/slow) Baseline XGBoost performance: R²=0.38, Accuracy=69% Data collected via GitHub REST API, November-December 2025.