Ensemble machine learning methods for crop yield prediction using adasvm and adanaive
An enhanced crop yield prediction solution combining AdaBoost with Support Vector Machine (AdaSVM) and Naive Bayes (AdaNaive) ensemble methods. Validated on multiple crops including rice paddy, cotton, sugarcane, groundnut, and black gram, the approach delivers higher accuracy and lower classification error than standalone SVM and Naive Bayes models for agricultural time series forecasting.
Overview
This solution offers an improved approach to crop yield prediction through ensemble machine learning techniques. By combining AdaBoost with Support Vector Machine (AdaSVM) and Naive Bayes (AdaNaive), the method addresses limitations of standalone classifiers and delivers more reliable time series forecasts for agricultural planning. Validation across multiple crops, including rice paddy, cotton, sugarcane, groundnut, and black gram, demonstrates consistent accuracy improvements over conventional SVM and Naive Bayes approaches, making it a practical tool for agricultural decision support.
Technical specifications
Core methodology:
- Ensemble classification combining AdaBoost with SVM (AdaSVM) and AdaBoost with Naive Bayes (AdaNaive)
- Time series analysis applied to historical crop yield datasets
- Data preprocessing through CSV import and attribute selection to refine input features
- Cross-validation operator used to train and test classification algorithms
- Implementation carried out using the RapidMiner data analysis platform
Validated performance results:
- SVM accuracy: 90.48% (rice paddy), 87.6% (cotton), 88.53% (sugarcane), 89.32% (groundnut), 86.7% (black gram)
- AdaSVM accuracy: 93.72% (rice paddy), 90.56% (cotton), 91.64% (sugarcane), 92.75% (groundnut), 89.42% (black gram)
- AdaSVM consistently outperforms standalone SVM across all tested crops
- Classification error correspondingly reduced with the AdaSVM approach
Technology readiness level
The solution has been validated through comparative testing on five crop datasets using established machine learning workflows in RapidMiner. Results confirm that AdaSVM and AdaNaive outperform baseline SVM and Naive Bayes classifiers in both accuracy and classification error metrics. The methodology is ready for further field validation and integration into broader agricultural forecasting systems, with potential for adaptation to additional crops and regional datasets.
About AREEO
AREEO is Iran’s national agricultural research, education, and extension organization, coordinating a comprehensive country‑wide system of institutes, provincial research and education centers, and field stations. Co‑located experimental farms and research stations enable rapid field validation and scale‑up, while an extensive extension network connects results directly to producers and agri‑industry. For commercialization, AREEO operates research incubation centers and supports knowledge‑based enterprises that engage with private partners. Research is primarily supported by national government programs through the Ministry of Agriculture Jihad, with additional competitive funding and international collaborations.
Inquire about this partnering listing
Looking for
Industry applications
delivered to your inbox.