How cadenbrannan2004-pixel/deadlock-win-probability works
DLWin is described as a Deadlock win-probability and build advisor, but the supplied README is empty and the sampled recommendation, feature, training, explanation, and service files provide no implementation details. The only evidenced workflow is manifest ingestion: configuration supplies the manifest URL and raw-data directory, ingestion fetches and saves the manifest, and prints a description. Other named modules are retained as product areas, but their connections and behavior cannot be established from the available excerpts.
Main entry points: ingest.py Manifest ingestion · config.py Runtime configuration · raw Raw manifest file
Components 10
Data ingestion
- Runtime configuration
src/dlwin/config.py - Manifest ingestion
src/dlwin/ingest.py - Raw manifest file
data/raw
Prediction and advice
- Feature engineering
src/dlwin/features.py - Win model training
src/dlwin/train_win_model.py - Build recommender
src/dlwin/recommender.py - Result explanation
src/dlwin/explain.py
Service interface
- Service application
src/dlwin/service/app.py - Live service
src/dlwin/service/live.py
Other components
- Deadlock data lake
Connections 3
- Manifest ingestion → Runtime configuration: reads settings
src/dlwin/ingest.py - Manifest ingestion → Deadlock data lake: fetches manifest
src/dlwin/ingest.py - Manifest ingestion → Raw manifest file: writes manifest
src/dlwin/ingest.py