Crop Yield Prediction using Systematic Review Process based Machine Learning Algorithm

Machine learning has become indispensable for predicting crop yields, significantly influencing crop selection and management strategies throughout the growing season. Extensive research has investigated various machine learning algorithms in this context. We conducted a systematic literature review (SLR) in order to compile and analyse the features and techniques used in these studies. A thorough search across six electronic databases identified 300 pertinent studies, of which 25 were meticulously selected for in-depth analysis according to specific inclusion and exclusion criteria. These studies were rigorously examined to assess the methodologies and features utilized, providing a foundation for future research recommendations. The analysis highlighted temperature, rainfall, and soil type as the most frequently used features in crop yield prediction, with one particular algorithm emerging as the most commonly applied. Convolutional Neural Networks (CNN) are the most common deep learning algorithms, followed by Long Short-Term Memory (LSTM) networks and Deep Neural Networks (DNN), according to 25 pertinent publications found via a second search that concentrated on deep learning-based research.