Thabang-Tsimakwane/botswana-food-inflation-forecasting

# Botswana Food Inflation Forecasting **Deep Learning IndabaX Botswana Hackathon 2024 — Phase 1 Submission** Forecasting Botswana's FAO food price index (item 23014, YoY % change) for January–December 2024 using two declared models: **Index SARIMA** (classical) and **LSTM** (deep learning). --- ## Quick Start ### Prerequisites - Python 3.10+ (tested on 3.12) - `git`, `make` (optional) ### Cross-Platform Setup **Bash / Linux / macOS / WSL / Git Bash:** ```bash python -m venv .venv source .venv/bin/activate pip install -r requirements.txt pytest ``` **PowerShell (Windows):** ```powershell python -m venv .venv .\.venv\Scripts\Activate.ps1 pip install -r requirements.txt $env:MPLBACKEND='Agg' $env:MPLCONFIGDIR="$PWD\tmp\matplotlib" pytest ``` ### Run Individual Models ```bash # Classical model — Index SARIMA (official submission) python train_sarima.py # → output/predictions.csv, output/sarima_results.json, output/sarima_backtests.csv # Deep-learning model — LSTM (candidate forecast) python train_lstm.py # → output/lstm_predictions.csv, output/lstm_results.json ``` ### Run Side-by-Side Comparison ```bash python run_pipeline.py # → output/model_validation_2023.csv, output/model_forecasts.csv # → output/figures/comparison_2023_models.png, residual_diagnostics.png, forecast_2024.png # → output/results.json (comparison metrics + HCP linkage) ``` ### Run Scenario Analysis (Phase 2) ```bash python run_pipeline.py # → output/scenario_hormuz.csv, output/scenario_sa_el_nino.csv # → output/scenario_baseline.csv, output/sensitivity_tornado.csv # → output/hcp_impact_scenarios.csv # → output/figures/scenario_report_p1_definitions.png ... p5_hcp_policy.png # → output/figures/scenario_report_5pages.pdf ``` **Scenarios simulated:** | Scenario | Shocks | Basis | |----------|--------|-------| | **Strait of Hormuz Closure** (mandatory) | BDI -30%, Brent +55% (via WB_Urea_yoy), FAO +18% (via WB_Maize_yoy) | 1990 Gulf War, 2019 Abqaiq, Ukraine wa …