YNF: Yield-Net Framework for Crop Yield Prediction Using Integrating Machine Learning Model

By providing food and work, the agricultural sector has greatly contributed to the growth of many nations across the world. However, figuring out how many crops are cultivated each year has shown out to be a significant challenge. In this work, we presented Yield-Net, a cutting-edge forecasting system that integrates machine learning methods including Gated Recurrent Unit (GRU), XGBoost, and Random Forest (RF). The output of crops from various places each year. The world agricultural yield dataset, which contains 28,242 entries with 7 variables for the years 1990 to 2013, was obtained from ourworldindata.org and is used in this experiment. The dataset was organized into a manner appropriate for modeling by performing data processing activities such managing missing values, removing duplicates, data encoding, and scaling. Training utilized 90% of the preprocessed data, whereas Yield-Net testing used 10%. Important characteristics in this study were chosen using XGBoost. The agricultural yield data was used to train the model, and GRU discovered sequential patterns in the chosen attributes. To predict the quantity of agricultural produce, Random Forest employs the characteristics gathered by GRU as input. The proposed YNF model has compared using XGBoost, LSTM, GRU, and RF model and trained over 25 iterations. The Root Mean Square Error (RMSE) and other cutting-edge metrics that are often used to evaluate the performance and degree of error of regression models were utilized during training. The Root Square Error (R2), Mean Absolute Error (MAE), and Square Error (RMSE) are used for evaluation. Extremely low performance inaccuracy is demonstrated by the Yield-Net framework, with an RMSE of 5.42 and an MAE of 3.95. At the same time, the proposed model R2 score shows 0.951 which indicates strong predictive capability of the proposed system on the unseen data. This paper contributes significantly to the ongoing research in the field of precision agriculture while laying a solid foundation for future work.