Prediksi Hasil Produksi Beras di Kabupaten Lamongan Menggunakan Stochastic Frontier Analysis (SFA)

The agricultural sector plays a crucial role in the Indonesian economy, especially in maintaining food estate and economic stability. This study aims to identify and improve the technical efficiency of rice production in Indonesia using Stochastic Frontier Analysis (SFA). Agricultural data were analyzed through validation, cleaning, feature selection, and modeling with a log-linear Cobb-Douglas production function estimated using Maximum Likelihood Estimation (MLE). Model performance was evaluated on training and test data using Log-likelihood, R-squared, and Mean Absolute Percentage Error (MAPE). The results showed good model prediction performance on test data (R-squared = 0.6658 and MAPE = 14.34%). Technical inefficiency analysis indicated that the level of inefficiency among farmers in Lamongan Regency was relatively low and homogeneous. However, the efficiency frontier analysis identified significant opportunities to increase rice yields through more efficient management of production factors and reduction of inefficiencies