Optimizing 5G Network Slicing Prediction in Next-Generation Networks Using CatBoost and Social Spider Algorithm

The rapid development of 5G networks has introduced the need for efficient network slicing to meet various service requirements, such as those in healthcare, smart cities, and IoT devices. Accurate prediction of network slices is crucial for managing these dynamic demands. In this context, this study proposes a system to improve the prediction of a 5G network slice using the CatBoost model with hyperparameters modified by the Social Spider Optimization (SSpiderO) algorithm. Collected from Kaggle, the dataset includes important metrics such as packet latency, slice type, and IoT use, which were balanced using the SMote method. Among numerous baseline models including SVM, Random Forest, and XGBoost, our model obtained a 0.9667 F1 score and 0.9668 precision. The strong predictive capacity of the model is shown by performance criteria including the ROC-AUC, precision, recall, and the F1 score. The results confirm the efficiency of SSpiderO in optimizing CatBoost in 5G network slicing, providing a strong solution for next-generation network management.