Measuring Poverty and Inequality with Reduced Data: A Machine Learning Approach Using Nigerian Household Data

In three linesStudy applying Random Forest Recursive Feature Elimination to Nigerian household survey data (2018/19) to identify minimal predictors of poverty status, welfare quintiles, and inequality. RF-RFE achieves 90% accuracy for poverty with 5 income variables, 80% for seasonal quintiles. ML methods reduce data requirements while preserving distributional information for poverty and inequality monitoring.

Summary generated by Claude — human-verified