Forecasting food price inflation in Nigeria and identifying its drivers using machine learning models

Abstract Persistent food price inflation has become one of Nigeria's most pressing macroeconomic and food security challenges, yet evidence on the most effective forecasting approaches and the principal drivers of food inflation remains limited. This study forecasts food price inflation in Nigeria over a two-year horizon and identifies its key determinants using monthly food price data. The analytical framework combines conventional time-series techniques with regularized econometric and machine learning models, including Autoregressive (AR), Ridge Regression, Least Absolute Shrinkage and Selection Operator (LASSO), Elastic Net, Partial Least Squares (PLS), and Artificial Neural Network (ANN). Time-series diagnostics confirmed that the food inflation series was stationary in levels, while the Akaike Information Criterion selected an AR (12) specification as the optimal benchmark model. Comparative out-of-sample evaluation revealed that the regularized regression models outperformed both the benchmark AR model and the nonlinear machine learning models. Elastic Net achieved the highest forecasting accuracy (RMSE = 9.94), followed by LASSO (RMSE = 9.95) and Ridge Regression (RMSE = 10.02), whereas PLS and ANN exhibited substantially poorer predictive performance. Feature selection results identified historical food prices as the most robust predictor of future food price inflation. Ridge Regression further highlighted exchange rate movements, pump prices, the food price index, and rainfall as important complementary predictors, underscoring the roles of inflation persistence, macroeconomic conditions, energy costs, and climatic factors in shaping food price dynamics. The findings demonstrate that parsimonious regularized regression models provide more accurate and interpretable forecasts than complex nonlinear algorithms in a structurally volatile macroeconomic environment. The study recommends strengthening inflation monitoring systems through regularized forecasting models while implementing coordinated macroeconomic, energy, and agricultural policies that improve domestic food production, enhance climate resilience, stabilize exchange rates, and reduce transportation costs to promote food price stability and food security in Nigeria.