Skip to main content

​
Install

git clone https://github.com/Idopt-Lab/airTravelModel.git
cd airTravelModel
python -m venv .venv
source .venv/bin/activate
pip install -e ".[test]"
Requires Python 3.10+.

​
Your first prediction

from air_travel_model import TransportPredictor

tp = TransportPredictor()

# Single mode — air fare, Q1 economy
fare = tp.predict("air_fare", "CHS", "ROA", quarter=1, is_business=False)
print(fare)
# {"price": 172.50, "currency": "USD"}

# All modes at once
results = tp.predict_all("CHS", "ROA", quarter=1)
for mode, result in results.items():
    print(f"{mode}: {result}")

​
Specifying locations

Locations can be provided three ways — all are equivalent: 
# IATA airport code
tp.predict("air_fare", "CHS", "JFK", quarter=2)

# City name (resolved to nearest airport)
tp.predict("air_fare", "Charleston, SC", "New York, NY", quarter=2)

# Lat/lon tuple
tp.predict("air_fare", (32.8986, -80.0406), (40.6413, -73.7781), quarter=2)
For train time, use Amtrak station codes for best accuracy: 
tp.predict("train_time", "CHS", "NYP", quarter=2)   # NYP = New York Penn
tp.predict("train_time", "WAS", "CHI", quarter=3)   # WAS = Washington, CHI = Chicago

​
Heuristic fallback

When a route has no training data, use estimate() instead of predict(): 
# estimate() always returns a value — never raises
result = tp.estimate("drive_time", "BFD", "UNV")
predict_all() and estimate_all() call all seven models in one shot.

​
Run the example script

python scripts/example.py
This demonstrates all modes on a Charleston SC → Roanoke VA route with annotated output.