Machine Learning for Crop Yield And Irrigation Energy Cost Prediction: Case Study of Five Tropical Crops

The agricultural sector plays a pivotal role in global food security and resource management. Accurate prediction of crop yields and energy costs is essential for improving agricultural productivity and sustainability. This study presents a novel approach that integrates machine learning techniques with a web-based application to predict crop yields and irrigation energy costs within the context of Ghana's farming sector. Comprehensive datasets containing historical crop yield records, rainfall, soil fertility status, and energy cost information were collected. Relevant features were extracted through a feature engineering process to improve prediction accuracy. Five machine learning algorithms - Linear Regression, Decision Tree, Random Forest, Feed-Forward Neural Network, and Recurrent Neural Network - were trained using the prepared datasets. Performance metrics such as Mean Absolute Error, Mean Squared Error, and R2 score were evaluated to assess the models' predictive capabilities and select the best model. The validation process confirmed the superiority of the Random Forest algorithm across the five major crops. A web-based application was developed for user-friendly interactions and real-time predictions. The application provides a user interface that allows stakeholders to access predictions and insights through web and mobile platforms. This study contributes to advancing agricultural technology by presenting a comprehensive framework for predicting crop yields and energy costs in Ghana's regions. By combining machine learning techniques with web applications, the approach offers accurate, scalable, and accessible predictions. This empowers stakeholders to make informed decisions and promote sustainable agricultural practices.