Identification of Cyber Bullying Using XGBoost Compared to Random Forest Classifier to Improve Accuracy

The study aims to improve the accuracy of cyberbullying detection. Compared to the Random Forest classifier, utilize XGBoost to improve accuracy. In this study, two groups were compared. The XGBoost Algorithm was compared in the study employing 1668 dataset samples, of which 573 were used for testing and 1554 for training. A sample size of 10 N was used for each in order to evaluate the detection of cyberbullying. The accuracy of the data is improved by the XGBoost with 94.416% and accuracy in comparison to Random Forest (87.703%). The detection of cyberbullying is statistically significant with a significant value of p=0.00192 (p<0.0496). Compared to Random Forest, the XGBoost Algorithm approach for detecting cyberbullying showed a notable improvement due to its increased accuracy.