A Machine Learning Approach to Address Food Security: Predicting Household Food Consumption in Bangladesh

This research investigates household food consumption habits in Bangladesh, highlighting their socio-economic consequences and predicting patterns with machine learning algorithms. A cleaned dataset including 209,264 items and 31 features was analyzed using data from the Food and Agriculture Organization (FAO), combined with data from the Intergovernmental Panel on Climate Change (IPCC) and other sources from 1990 to 2021. Critical elements like urbanization, agricultural methodologies, and demographic trends were analyzed for their impact on food consumption. Six supervised machine learning models- Linear Regression (LR), Ridge Regression (RR), Support Vector Regression (SVR), Decision Tree Regression (DTR), Random Forest Regression (RFR), and Polynomial Regression (PR) were employed, and the models were evaluated using evaluation metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R Squared (R2). The results indicate that RFR and LR are the most effective models, achieving an R2 of 0.99, representing greater prediction accuracy than other approaches. This study outperforms prior findings using advanced preprocessing methods and strong modeling frameworks. The findings highlight the effectiveness of predictive modeling in policy formulation, providing insights to address food security issues and inflationary patterns in Bangladesh. These findings establish a foundation for enhancing dietary diversity, urban food systems, and sustainable agriculture methods in developing economies.