The h-index Metric for GitHub
Summary
A bachelor's thesis that adapts the academic h-index to measure developer influence across 3.5M GitHub users using machine learning.
The Problem
The h-index is a metric used in academia to measure research impact. I adapted it for GitHub to answer: can we measure a developer's influence? No standard metric existed, so I built one from scratch.
The Solution
Collected data on 3.5M users and 30M repositories using the GitHub GraphQL API. Stored and queried data with SQLite. Built a Random Forest model with Scikit-Learn and Pandas to predict developer influence scores. Trained and evaluated the model on separate datasets to ensure accuracy.
Presented this project at Študentská vedecká konferencia and was published.
Key findings from the data:
GitHub h-index distribution:
Top 5 developers by h-index score:
