ADVANCED DATA ANALYSIS AND CLUSTERING METHODS FOR CROP YIELD PREDICTION USING ARTIFICIAL SYSTEM



Authors

  • 1Dhara Jaya Soniya 2Dr. Prasuna Grandhi 3Dr. A. Tirupataiah

DOI:

https://doi.org/10.15282/jmes.17.1.2023.10.0759


Keywords:

Agriculture, Crop Prediction, Machine Learning, Deep Learning, SVM, LSTM, RNN, Suitability Assessment, Temporary Crops, K Means, Machine Learning, Prediction


Abstract

The science and skill of cultivating plants and fauna are referred to as agriculture. 60.45% of Indian land is used for farming, which gives the country the second-highest global ranking. These agricultural economy-related problems result in higher crop yield. Crop Yield Prediction is crucial in today's agriculture market (CYP), which is expanding quickly. Selected features and machine learning techniques are necessary for accurate prediction. An electronic agricultural record (EAR), which is effectively recommended by agronomists as a vital component of a smart crop system, is intended to integrate many different datasets. Dirichlet Allocation and Artificial Neural Network classification algorithms are used to determine the causes of a given plant disease. To predict a suitable yield, the climate and soil conditions are taken into account.While LSTM and RNN are employed as Deep Learning algorithms, the SVM is implemented as a Machine Learning method. There are numerous clustering techniques, including k-Means, Expectation-Maximization, Hierarchical Micro Clustering, Density-Based Clustering, and Weight-based Clustering, which are briefly discussed. A novel clustering method, Epsilon Density-Based Prediction (EDBP), is also suggested to update the best crop production prediction. In order to anticipate yield, this paper uses ANN with cascade-forward back propagation and Elman back propagation. We employed Regression methods, Decision trees, Naive Bayes, SVM, K-Means, Expectation-Maximization (EM), and AI approaches (LSTM, RNN) together with machine learning and deep learning algorithms. For predicting agricultural yield the Random Forest algorithm has a training accuracy of 99.27%. With 95% accuracy and 92% sensitivity, the suggested system produces good results.



Published

2024-06-30

How to Cite