Abstract
Black gram, a pulse rich in protein, is mostly cultivated during Rabi in Andhra Pradesh, helping smallholder growers and improving soil fertility. The area, production, and productivity of Black gram can be predicted based on historical data that can provide potential insights. However, historical methods often do not work because they ignore recent shifts in climate, pest status, and responses in the market. This manuscript proposes a gaussian process regression (GPR), random forest (RF) named as GPRF method for forecasting the area, production, and productivity of black gram in Andhra Pradesh using historical agricultural data. Initially, input data is collected from India stat dataset. To execute, the input data is pre-processed using GPR, which removes null values from the data in the dataset. Then the preprocessed data is provided to RF, which is employed to forecast the area, production and productivity for rainfed crop black gram. The approach achieved a lower MAE of 0.028 and an RMSE of 0.021, outperforming existing techniques such as SVR, KNN and SVM. The forecasting period spans from 2026 to 2030 and focuses on area, production, and productivity. The proposed GPRF model accurately forecasts black gram cultivation in Andhra Pradesh, with low MAPE values: 1.33% for area, 2.04% for production, and 0.73% for productivity. The results indicate a strong upward trend, and the proposed method outperforms existing methods in terms of accuracy and reliability, making it highly suitable for strategic agricultural planning and policy-making.
Keywords: Agricultural Economics, Black Gram, Climate Variability, Crop Productivity Forecasting, Precision Agriculture, Rainfed Crops.