TransPort-PH: Transportation Policy Analysis for the Philippines

TransPort-PH is a machine learning and causal inference framework for analyzing transportation policies, forecasting congestion, and evaluating economic impacts in the Philippines and globally.

🚀 Key Features

• Forecasting: Temporal Fusion Transformer (TFT) for multi-horizon time series predictions.
• Causal Analysis: DoWhy framework for robust policy impact assessment.
• Simulation: Counterfactual scenarios for "what-if" analysis.
• Interactive Dashboard: Streamlit-based tool for visualization.
• Rich Data: Integrates World Bank, DPWH, and other sources (7,430+ observations).

⚡ Quick Start

Installation

git clone https://github.com/Acteus/TransPort-PH.git
cd TransPort-PH
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt

Usage

Run Dashboard (Visualization & Analysis)

streamlit run src/visualization/dashboard_app.py

Run Full Pipeline (Data Collection -> Analysis)

python src/utils/run_all.py

📂 Project Structure

• src/: Main source code (data collection, models, analysis, visualization).
• data/: Processed datasets (World Bank, etc.).
• config/: Configuration settings.
• docs/: Detailed documentation.

📊 Data & Methodology

Sources: World Bank, DPWH, JICA, LTFRB, PSA, OpenAQ, TomTom.

Methods:

• ML: Temporal Fusion Transformer (TFT)
• Causal: DoWhy (Treatment Effect Estimation)
• Validation: Sensitivity analysis & baseline comparisons (ARIMA, LSTM)

Key Results:

• Significantly expanded data coverage (275 regions).
• TFT model achieves 0.24 validation loss, outperforming baselines.
• Actionable insights on transit investment vs. congestion.

📚 Documentation

• Quick Start Guide
• Dashboard Guide

📄 License & Citation

MIT License. See LICENSE.

@software{transport_ph_2024,
  title={TransPort-PH: Transportation Policy Analysis for the Philippines},
  author={[Your Name]},
  year={2024},
  url={https://github.com/Acteus/TransPort-PH}
}