A Machine Learning Model for Predicting Food Price Variations Under Currency Fluctuation Scenarios in Afghanistan

In developing nations, particularly countries that depend significantly on imports like Afghanistan, food price volatility continues to pose a serious threat to both economic stability and food security. By using the (WFP) Food Prices dataset for Afghanistan, this study presents a machine learning based system for forecasting changes in food prices under currency fluctuation scenarios. The models reflect the implicit effects of exchange rate fluctuations on domestic food markets by combining pricing data denominated in Afghani (AFN) and U.S. dollars (USD). Three algorithms from ML: Linear Regression, Random Forest, and Long Short-Term Memory (LSTM) networks, were trained and evaluated using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and the Coefficient of Determination (R²). With an MAE of 3.2, RMSE of 4.3, and R² of 0.89, the Random Forest model outperformed both LSTM (MAE 3.8, RMSE 4.9, R² 0.86) and Linear Regression (MAE 4.5, RMSE 5.8, R² 0.82). The findings show a strong relationship between market inflation and exchange rate volatility, indicating that currency depreciation significantly increases domestic food prices. The study emphasizes the value of machine learning in creating early warning systems and policy tools that enhance food price stability and support food security in poor countries like Afghanistan.