This thesis focuses on the problem of short-term economic forecasting with very limited data because of russia’s invasion of Ukraine. The invasion has created a gap in the release of statistical data, making traditional forecasting models less efficient, particularly in the regional forecasting scope. The study uses RMSE, CRPS, and SMAPE metrics to evaluate the forecast’s accuracy while measuring accuracy against traditional approaches. The results show that strong nowcasting models can be developed even with sparse data. This work contributes to the literature on economic monitoring during crises and provides a framework for real-time forecasting in post-conflict reconstruction.