Comparison of Various Text Classification Methods for Gadget User Sentiment at an Early Age

Abstract

In the context of rapid digital development, the use of gadgets among Indonesian children has become a very important topic to study. This study aims to analyze sentiments related to gadget use by applying classification methods such as Support Vector Machine (SVM), Naïve Bayes, and Decision Tree. To overcome data imbalance, After applying the SMOTE technique, the results of the study revealed that SVM obtained the highest accuracy of 99% with SMOTE, followed by Decision Tree which reached 98% and Naïve Bayes which obtained 94% when SMOTE was applied. In addition, the application of preprocessing techniques such as tokenization, stemming, and filtering contributed to improving data quality. These findings emphasize the importance of choosing the right method in sentiment analysis to understand the impact of gadget use on children's development. This study provides meaningful insights for the development of better policies and practices related to children's digital device use