Uncovering Community and Project Insights Through Data Driven Methods
summary
Shows a working pipeline that pulls GitHub PRs with the mi/thoth collector, stores features in Trino and S3, and trains a random-forest classifier to predict PR merge intervals (e.g. 15–24 hours class). Models are packaged with a scikit-learn pipeline, served with Seldon in the Operate First cloud, and wired into a GitHub bot that posts predictions on new PRs; demo and dashboards run end-to-end and deliver HTTP 200 responses with class labels.
AI-written summary · about
chapters (10)
- 0:00Start
- 2:45How Can You Use Ai or Machine Learning To Aid a Project's Development
- 3:39Project Ai for Ci
- 12:09Dashboard
- 19:38Machine Learning Setup
- 20:14Creating the Model
- 22:11Model Deployment
- 24:30Github Bot
- 26:30Demo
- 31:46Next Steps
related talks
24mUncovering Community and Project Insights Through Data Driven Methods
31mTowards an open operations data community
29mExtract, Transform and Learn About Your Developers
34mData pipelines powered by Open source and fun
36mBuild prod ML systems with feature /training /inference pipelines
57mUsing developer-centric data to predict, prioritize, and improve App Security Outcomes
05mData Science in DevOps/SysOps
45mGitHub Archive - Processing and analyzing the billions of events generated by the w