Skip to content

⚙️ Configure Project Paths and Options

Runtime settings are centralized in YAML files, defining project paths, data scope, and feature flags per pipeline.

📂 Configuration Files

All configuration settings are defined per pipeline. Edit the file corresponding to your use case:

  • 📡 OpenSky 2025: scripts/opensky/config.yaml
  • ✈️ QAR (Private): scripts/qar/config.yaml

📝 Configuration Structure Example

This example shows the primary fields in the OpenSky configuration.

scripts/opensky/config.yaml
paths:
  data_dir: "/path/to/data"
  download_dir: "downloaded_parquet"
  preprocess_dir: "preprocessed_parquet"
  # ... (more directories)
  era5_cache_dir: "era5_cache"

era5_features:
  - u_component_of_wind
  - v_component_of_wind
  - temperature

typecodes:
  - A320
  - A20N
  # ...

bada:
  bada_4_2_dir: "/path/to/BADA/4.2.1"

🔑 Key Parameters and Best Practices

Section Parameter Type Best Practice / Description
Paths data_dir Path Crucial: Keep this path absolute. All subfolders (download_dir, models_dir, etc.) are resolved relative to this root.
Paths era5_cache_dir Path Path for local cache of meteorological fields. Setting this prevents re-downloading large files.
Scope typecodes List Single Source: Adjust aircraft type scope here, not by modifying pipeline scripts.
BADA bada_4_2_dir Path Set this only if you plan to run baseline evaluation (07_bada_prediction.py).
ERA5 era5_features List Defines the specific meteorological fields (wind components, temperature) to be used as exogenous inputs.

QAR Pipeline Variations

The QAR configuration is minimal. It typically only retains essential paths (data_dir, predicted_dir, models_dir), typecodes, and options for parallel processing (computing.default_cpu_count).

Directory Existence

Ensure your main directories exist before running data downloads or preprocessing scripts. 

mkdir -p /path/to/data/

🚀 Next Steps