Implementation of Logistic Regression for Cyberbullying Classification on Indonesian Instagram Comments

Abstract

The rapid growth of social media usage, especially Instagram, has increased user interaction while also raising the occurrence of cyberbullying in the form of insulting, mocking, and offensive comments that may negatively affect victims’ psychological conditions. This study aims to develop a cyberbullying detection model for Indonesian-language Instagram comments using the Logistic Regression algorithm with a Natural Language Processing (NLP) approach. The dataset used consists of 650 comments labeled as cyberbullying and non-cyberbullying. The preprocessing stages include cleaning, case folding, tokenization, stopword removal, and stemming, followed by text transformation into numerical representation using the Bag of Words method with CountVectorizer. The research applies the CRISP-DM methodology consisting of business understanding, data understanding, data preparation, modeling, evaluation, and deployment. The evaluation results show that the Logistic Regression model performs well in classifying comments, achieving an accuracy of 83%, precision of 0.83, recall of 0.83, and F1-score of 0.83. These findings indicate that the combination of the Bag of Words method and Logistic Regression algorithm is effective for detecting cyberbullying in Indonesian Instagram comments.