Accurate Prediction of Pest Information Patterns in Crops Using Machine Learning Algorithms

Aim: The objective of this study is to create an effective pest prediction model utilizing Support Vector Machine (SVM) and assess its performance against the Decision Tree model. The models evaluate environmental factors such as temperature, humidity, soil moisture, and pest population density to improve the accuracy of pest forecasting. Materials and Methods: This research is divided into two groups. Group 1 applies the Decision Tree model, which is trained with 26 samples, to classify pest infestations based on environmental factors. Group 2 utilizes the SVM model, which is trained with 26 samples, employing kernel functions for enhanced classification accuracy. The G Power value is set at 80%, with a threshold of 0. 05% and a confidence interval of 95%. Significance: < 0.05 Result: The SVM model reached greater accuracy (85% to 92%) when compared to the Decision Tree model (75% to 85%). The precision of the SVM model varied from 0. 83 to 0. 91, whereas the precision of the Decision Tree model ranged from 0. 80 to 0. 83. The best classification accuracy was noted at 90% for SVM, with a statistical significance of p. Conclusion: The research finds that the SVM model greatly surpasses the Decision Tree model in pest forecasting, offering improved accuracy, precision, and recall. Subsequent studies may investigate hybrid models and the incorporation of real-time sensors to further improve predictive accuracy for sustainable pest control.