Skip to content

About

Predicts global food supply chain disruptions by processing 300k+ climate, economic, and geopolitical data points via an Airflow ETL pipeline. Utilizes LSTM (TensorFlow) and ARIMA (Statsmodels) models, with automated data validation and hyperparameter tuning. Features a real-time Streamlit dashboard for monitoring and visualization.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

2 Commits

Folders and files

Repository files navigation

Global Food Supply Chain Resilience Analyzer

GitHub

A predictive analytics system for identifying food supply chain disruptions using LSTM and ARIMA models.

Repository

Features

  • ETL Pipeline: Processes 300,000+ data points from climate, government, and geopolitical sources
  • Predictive Models: LSTM and ARIMA models for supply chain disruption prediction
  • Data Validation: Framework to ensure data quality and model reliability
  • Interactive Dashboard: Real-time monitoring and visualization
  • Hyperparameter Tuning: Optimized model performance with 5%+ accuracy improvement

Project Structure

food_supply_chain_analyzer/
├── data/                   # Data storage
├── src/                    # Source code
│   ├── etl/               # ETL pipeline
│   ├── models/            # ML models
│   ├── validation/        # Data validation
│   └── dashboard/         # Streamlit dashboard
├── notebooks/             # Jupyter notebooks
├── tests/                 # Unit tests
└── config/                # Configuration files

Quick Start

  1. Install Dependencies

    pip install -r requirements.txt
  2. Run the Dashboard

    streamlit run src/dashboard/main.py
  3. Run ETL Pipeline

    python src/etl/pipeline.py
  4. Train Models

    python src/models/train_models.py

Data Sources

  • Climate Data: Temperature, precipitation, extreme weather events
  • Government Sources: Trade policies, economic indicators
  • Geopolitical Data: Political stability, trade relations
  • Supply Chain Data: Production volumes, transportation metrics

Model Performance

  • Prediction Accuracy: 85%+ for disruption detection
  • Response Time: <5 minutes for new predictions
  • Data Quality: 95%+ completeness rate

Technologies Used

  • Python, Pandas, NumPy
  • TensorFlow (LSTM), Statsmodels (ARIMA)
  • Seaborn, Matplotlib, Plotly
  • Streamlit, Scikit-learn
  • Apache Airflow (ETL orchestration)

Model & Data Access

All trained model files (.h5, .pkl) and processed data (.csv) are available on Hugging Face:

Downloading Models & Data

You can download files directly from the Hugging Face web interface or programmatically using the huggingface_hub Python package:

from huggingface_hub import hf_hub_download

# Example: Download a model file
model_path = hf_hub_download(
    repo_id="IshaanPotle27/global-food-supply-chain-models-and-data",
    filename="models/arima_temperature_celsius_forecaster.pkl",
    repo_type="dataset"
)

# Example: Download a data file
csv_path = hf_hub_download(
    repo_id="IshaanPotle27/global-food-supply-chain-models-and-data",
    filename="data/food_supply_chain_data.csv",
    repo_type="dataset"
)

License

MIT License

About

Predicts global food supply chain disruptions by processing 300k+ climate, economic, and geopolitical data points via an Airflow ETL pipeline. Utilizes LSTM (TensorFlow) and ARIMA (Statsmodels) models, with automated data validation and hyperparameter tuning. Features a real-time Streamlit dashboard for monitoring and visualization.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages