Predictive Modelling of Cattle Market Prices in Maiduguri Using Regression Analysis

Cattle market price volatility in North East Nigeria affects farmer incomes, food security, and regional trade, and predictive modelling of price movements supports smallholder livestock producers in making informed market participation decisions. This study developed regression models for weekly cattle price forecasting at Maiduguri's Mile 12 livestock market, the largest cattle market in the Lake Chad Basin region. Weekly price records from January 2020 to December 2022, obtained from Borno State Ministry of Agriculture, were merged with rainfall data, conflict incidence indicators from ACLED, and seasonal calendar variables. Multiple linear regression, elastic net regularisation, and gradient boosting regression were compared. Gradient boosting achieved the lowest RMSE of N18,400 per head and an R-squared of 0.81 on test data. Conflict incidence in preceding weeks was the strongest negative predictor of price levels, followed by rainfall-driven pasture availability and proximity to Islamic calendar festive periods. The model correctly identified the January 2021 and December 2021 price spikes associated with Eid celebrations. Coefficient analysis confirmed that prices were significantly higher in drought-affected quarters. Elastic net regularisation provided useful variable selection, reducing the feature set from 24 to 11 variables without significant accuracy loss. Recommendations include making weekly price predictions available to farmers through SMS-based advisory services, developing a Maiduguri livestock market price app, and expanding the monitoring framework to include Gashua and Potiskum markets to improve regional market coverage.

Keywords: cattle market price prediction, gradient boosting, Maiduguri, livestock analytics, North East Nigeria