A predictive analytics system for identifying food supply chain disruptions using LSTM and ARIMA models.
- 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
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
-
Install Dependencies
pip install -r requirements.txt
-
Run the Dashboard
streamlit run src/dashboard/main.py
-
Run ETL Pipeline
python src/etl/pipeline.py
-
Train Models
python src/models/train_models.py
- 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
- Prediction Accuracy: 85%+ for disruption detection
- Response Time: <5 minutes for new predictions
- Data Quality: 95%+ completeness rate
- Python, Pandas, NumPy
- TensorFlow (LSTM), Statsmodels (ARIMA)
- Seaborn, Matplotlib, Plotly
- Streamlit, Scikit-learn
- Apache Airflow (ETL orchestration)
All trained model files (.h5, .pkl) and processed data (.csv) are available on Hugging Face:
- Hugging Face Dataset Repo: IshaanPotle27/global-food-supply-chain-models-and-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"
)MIT License