Forecasting monthly incident trends across global regions using machine learning.
📘 About the Model
This dashboard analyzes data from the Global Terrorism Database (GTD) and
forecasts the expected number of incidents per region for upcoming months.
The model uses LightGBM and is tuned per-region.
🧠 Lag and rolling features (trend, volatility, zero-share).
🌍 Trained separately for 12 world regions.
📈 Evaluated with SMAPE, MAE, and RMSE.
Note: Model reliability differs across regions. Interpret forecasts in conjunction with the
validation metrics (MAE, RMSE, SMAPE, and correlation), which quantify
out-of-sample predictive accuracy.