Forecasting Demand for Agricultural Inputs Using Time Series Methods in Sokoto State

Accurate demand forecasting for agricultural inputs such as fertilizers, improved seeds, and pesticides is essential for effective supply chain planning in Nigeria's agrarian northern states. This study applies time series forecasting methods, including moving averages, exponential smoothing, and the Holt-Winters seasonal model, to project quarterly demand for NPK fertilizer in Sokoto State, North West Nigeria. Secondary data spanning 10 years were obtained from the Sokoto State Agricultural Development Programme, covering quarterly distribution volumes across 23 local government areas. The three methods are fitted to historical data, and forecast accuracy is evaluated using mean absolute deviation (MAD), mean squared error (MSE), and mean absolute percentage error (MAPE). The Holt-Winters model consistently outperforms the simpler methods across all accuracy metrics, capturing both trend and seasonal patterns associated with planting seasons and government subsidy cycles. Demand projections for the subsequent four quarters are generated and benchmarked against planned procurement volumes. Results reveal a supply-demand gap of approximately 18,000 metric tonnes per quarter during peak planting season, suggesting systematic under-procurement. The study recommends institutionalizing Holt-Winters forecasting within the State's input procurement planning cycle and building a shared demand database accessible to input dealers and government agencies. This research contributes a replicable forecasting framework for agricultural input supply management across Nigeria's Northwest zone. Keywords: time series forecasting, demand analysis, agricultural inputs, Sokoto State, Holt-Winters model