HYBRID CS-XGBOOST: REVOLUTIONIZING TOMATO DISEASE PREDICTION FOR IMPROVED AGRICULTURAL YIELD AND QUALITY

Abstract

In agricultural informatics, the accurate prediction of tomato diseases is crucial for optimizing yield and maintaining quality. This study introduces an innovative hybrid algorithm that synergistically combines the meta-heuristic Cuckoo Search (CS) with the gradient boosting capabilities of XG Boost. The proposed model aims to predict five distinct states of tomato health: No Disease, Early Blight, Late Blight, Leaf Mold, and Tomato Yellow Leaf Curl Virus. By fusing CS's prowess in optimized feature selection with XG Boost's robustness in classification, the hybrid model endeavors to enhance the predictive precision. A comparative analysis was conducted against benchmark algorithms, namely KNN, SVM, Random Forest, standalone XG Boost, and Cat Boost. Preliminary results, evaluated based on standard metrics like accuracy and F1-score, indicate that the hybrid CS-XG Boost algorithm manifests a marked improvement in prediction accuracy and computational efficiency. This research underscores the potential of integrating meta-heuristic search algorithms with gradient boosting models, providing a new avenue for advancements in agricultural disease prediction.

Keywords

• tomato disease prediction
• cuckoo search (cs)
• xgboost hybrid model
• meta-heuristic integration
• meta heuristic integration