Potato Late Blight Outbreak: A Study on Advanced Classification Models Based on Meteorological Data

Area of Science:

• Agricultural Science
• Plant Pathology
• Machine Learning

Background:

• Late blight infection detection is important, but predicting outbreaks is key for economic potato production.
• Minimizing pesticide use is vital for human health and environmental safety.

Purpose of the Study:

• To develop a predictive model for potato late blight outbreaks.
• To enhance potato crop management and reduce economic losses.

Main Methods:

• Utilized real-time European data from 1980-2000 for precise late blight classification.
• Incorporated hybrid machine learning models, including a stacking classifier and logistic regression.

Main Results:

• Achieved a highest prediction accuracy of 87.22% for potato late blight outbreaks.
• Demonstrated the effectiveness of hybrid models in forecasting plant disease.

Conclusions:

• Predictive modeling of late blight is crucial for efficient potato health management.
• Further model enhancements and data integration can improve prediction accuracy and reduce production costs.