wku572/teff-price-dss

Deep learning-based decision support system for forecasting teff market prices in Ethiopia # Teff Market Price Decision Support System A Streamlit-based Decision Support System (DSS) for forecasting weekly retail teff prices in Ethiopia using a trained Long Short-Term Memory (LSTM) deep learning model. --- ## Features - One-week-ahead retail teff price forecasting - Hierarchical selection of Region, Zone, Market, and Variety - Automatic loading of the latest 12 weeks of historical observations - Editable historical inputs for what-if analysis - Interactive forecast visualization - Model performance comparison (LSTM, GRU, MLP, and TCN) - Forecast history management with CSV export --- ## Prerequisites - Python 3.12 - Git --- ## Installation ### 1. Clone the Repository ```bash git clone github.com cd teff-price-dss ``` ### 2. Create a Virtual Environment **Windows** ```bash python -m venv .venv .venv\Scripts\activate ``` **Linux / macOS** ```bash python3 -m venv .venv source .venv/bin/activate ``` ### 3. Install Dependencies ```bash pip install -r requirements.txt ``` --- ## Required Files The following research artifacts are **not included** in the repository and must be copied into the project before running the application. ### Dataset Place the dataset in: ```text data/ └── Final_Cleaned_Teff_Dataset_2020-2026.xlsx ``` ### Model Artifacts Place the trained model and preprocessing artifacts in: ```text artifacts/ ├── best_lstm_model.pt ├── sequence_scaler.joblib ├── target_scaler.joblib ├── static_feature_encoder.joblib └── model_comparison.csv ``` --- ## Run the Application Start the Streamlit application: ```bash streamlit run Home.py ``` Then open your browser and navigate to: ``` localhost ``` --- ## Verify the Application The application should provide the following pages: - Home - Forecast - Model Performance - Forecast History - About From the **Forecast** page you can: 1. Select Region 2. Select Zone 3. Select Market 4. Select Teff Variety 5. Automatically load t …