How andrew-canada/flightid works
FlightID is a batch airfare-audit pipeline: it turns airline data into fare-per-mile predictions, inflation flags, explanations, alternative destinations, and corridor rankings. The main workflow is implemented in src/train_explain.py, with feature engineering and analysis helpers in src/utils.py; generated artifacts are written to outputs/. The README also describes notebook-based visualizations. Edges reflect the sampled pipeline and documented capabilities; notebook internals were not sampled.
Main entry points: train_explain.py Pipeline runner · outputs Analysis outputs · airline_ticket_dataset_csv.csv Airline dataset · visuals.ipynb Notebook visualizations
Components 12
Airfare inputs
- Airline dataset
data/airline_ticket_dataset_csv.csv
Analysis pipeline
- Pipeline runner
src/train_explain.py - Feature engineering
src/utils.py - Fare model
src/train_explain.py
Fare insights
- SHAP explanation
src/utils.py - Inflation detection
src/utils.py - Alternative routes
src/utils.py - Corridor rankings
src/utils.py - Notebook visualizations
notebooks/visuals.ipynb
Reports and artifacts
- Model artifacts
outputs - Analysis outputs
outputs
Other components
- Pipeline operator
Connections 11
- Pipeline operator → Pipeline runner: runs
README.md - Airline dataset → Pipeline runner: loads
src/train_explain.py - Pipeline runner → Feature engineering: engineers features
src/train_explain.py - Pipeline runner → Fare model: trains
src/train_explain.py - Pipeline runner → SHAP explanation: computes explanations
src/train_explain.py - Pipeline runner → Inflation detection: detects inflation
src/train_explain.py - Pipeline runner → Alternative routes: suggests routes
src/train_explain.py - Pipeline runner → Corridor rankings: ranks corridors
src/train_explain.py - Fare model → Model artifacts: saves model
src/train_explain.py - Pipeline runner → Analysis outputs: writes reports
src/train_explain.py - Notebook visualizations → Analysis outputs: visualizes results
README.md