An Innovative Hybrid Approach for Predicting Crop Yields across Diverse States and Crops

Abstract

The use of machine learning and deep learning may help estimate crops by automatically extracting attributes and learning from information. Meanwhile, smart farming technology aids farmers in increasing yields by isolating key factors in plant development. In this study, we present a unique hybrid approach (SLR) for predicting crops by extracting key characteristics from SVM (Support Vector Machine), LSTM (Long-Short Term Memory), or RNN (Recurrent Neural Network).  Here, we experimented with crops from a variety of states to estimate their potential yields in quintals per hectare. The suggested method has a37.45 percent precision rate, an 80.00 percent recall rate, a51.01 percent fscore, a93.64 percent specificity rate, and an accuracy rate of 93.02 percent.