Optimizing Crop Yield Prediction using Machine Learning and Deep Learning Approach

Agriculture serves as the backbone of many nations, including India, where it holds the second position globally in production. Contributing significantly to the country’s GDP, Indian agriculture’s vitality cannot be overstated. Early prediction of crop yields proves pivotal in enabling farmers to optimize production rates and strategize effectively. In our research, we employed Machine Learning (ML) & Deep Learning (DL) techniques to forecast crop yields for Bihar, India, leveraging publicly accessible datasets. Various algorithms were evaluated, including Random Forest, Linear Regression, K-Nearest Neighbour (KNN), Support Vector Machine (SVM), Keras Nadam Optimization, Long Short-Term Memory (LSTM), and Artificial Neural Network (ANN). Among these, LSTM emerged as the top performer, boasting a Mean Absolute Error (MAE) of 0.016, Root Mean Square Error (RMSE) of 0.086, and achieving an accuracy of up to 98.311%. The utilization of LSTM demonstrates its efficacy in capturing complex temporal relationships within the data, thereby facilitating highly accurate crop yield predictions. Such advancements in predictive modelling hold immense promise for enhancing agricultural outcomes and supporting the livelihoods of farmers.