Agriculture plays a crucial part in the economy of many countries, particularly in de-veloping nations, where farming serves as the primary source of income for many in-dividuals. This study aims to enhance agricultural productivity through three key areas: yield prediction, price forecasting, and crop recommendation. The study utilizes remote sensing data and five distinct datasets to analyze agribusiness predictions. Following that, assess the performance robustness using new datasets, statistical analysis, and model parameters, and compare the results with earlier studies. The research fol-lows three main approaches: (1) Hybrid Technique for Yield Prediction: This method employs the Gradient Boosting Regressor, Cat Boost Regressor, and Bagging Regressor, utilizing the ensemble aggregation technique of stacking to predict crop yield. (2) Combined multinomial logistic regression and Yeo-Johnson transformer for Crop Recommendation: A logistic transformer is utilized to recommend appropriate crop varieties based on remote sensing data, aiming to improve precision agriculture. (3) Deep Neural Network for Price Forecasting: The study applies deep neural network models, especially a Multi-Layer Perceptron (MLP), to predict future crop market prices. Using MLP in price forecasting is a novel application in this context. Moreover, the food price data is evaluated using traditional algorithms, such as the Prophet and Auto-regressive Integrated Moving Average (ARIMA), to further validate the prices. Lastly, the study compares these approaches with previous research and conventional models to evaluate their effectiveness. Following that, LIME outlines model results alongside transparency, insight, and trust. Then, real-time deployment allows for immediate predictions. Last but not least, agribusiness integration corresponds to performance, business impact, and workflow. The ultimate goal of this forecast analysis is to meet the growing demand in the agricultural sector by providing farmers with tools for complete crop prediction. In the future, we intend to create a mobile application that will give users access to a multitude of data, making it simpler to forecast various crops for the agricultural market at every moment.
Khin et al. (Tue,) studied this question.