Integrating Machine Learning Models for Chili Disease Prediction and Economic Impact Analysis

In this research, we seamlessly integrate machine learning methodologies into chili farming practices within tunnel environments, addressing the dual challenges of disease detection and market trend interpretation. Our study employs an image processing algorithm enhanced with machine learning techniques, achieving high accuracy in identifying and diagnosing prevalent chili crop diseases in Sri Lanka. With an impressive F1 score of 0.9444 and a mean absolute error of 0.1489, a training accuracy of 98.58% our disease prediction model provides practical mitigation strategies to farmers, ensuring minimal losses. 50 instances were separated from each class for testing purposes. 80% of the remaining data was utilized for training, while the remaining 20% was allocated for validation. Simultaneously, we leverage historical market data and price trend analysis through a mobile application, enabling real-time prediction of the most profitable chili crop varieties for cultivation. However, our findings highlight challenges in market trend analysis, reflected in our training MAE of 7.57 and RMSE of 16.54. Although our disease prediction model excels, the integration of both concepts remains a challenge. Despite this, our research lays the foundation for transformative impacts on chili farming, propelling productivity, minimizing losses, and fostering sustainable growth in the industry.