Crop yield prediction using multi-attribute weighted tree-based support vector machine

• Published in: Measurement. Sensors 2024-02, Vol.31, p.101002, Article 101002
• Main Authors: Rajakumaran, M., Arulselvan, G., Subashree, S., Sindhuja, R.
• Format: Article
• Language: English
• Subjects: Agriculture; Crop productivity; Genetic algorithm (GA); Machine learning (ML); Principal component analysis (PCA); Z-Score normalization
• ISSN: 2665-9174; 2665-9174
• DOI: 10.1016/j.measen.2023.101002

Agriculture is a prevalent and low-paying occupation in India. Agriculture is the key factor in the development of civilization. India is a predominantly agrarian nation with a crop-based economy. We can argue that agriculture can sustain the economy of our nation as a consequence. When organizing an agricultural enterprise, each crop must be properly picked. Numerous variables, such as market price, production volume, and governmental restrictions, will affect the crop selection. Numerous changes need to be made in the agriculture sector if economic growth in India is to be improved. The need for an effective technique to simplify crop cultivation and aid farm owners in crop management and production arises from the environment's unpredictable climate variables. This might enable potential farmers to practice sustainable agriculture. By modifying the economic position by producing the optimal crop, machine learning (ML) may lead to a revolution in the field of agriculture. To accurately predict crop yield, this research proposes the multi-attribute weighted tree-based support vector machine (MAWT-SVM) approach. Python software was used to conduct this study. Due to the existence of noisy data, the dataset for this study is first obtained, and the obtained raw data is normalized using the z-score normalization approach. Then, to extract the important features, principal component analysis (PCA) being employed. Additionally, the genetic algorithm (GA) employed to improve the suggested MAWT-SVM's performance by selecting the best features. Contrasted with other methods, the solution show that the suggested method achieves the best performance in crop yield prediction.