Sentiment Analysis of Thesis Policy by Instagram Users Using Logistic Regression
Fathia Dwi Astuti, Widodo Widodo, Bambang Prasetya Adhi · 2024
The requirements for university graduation were changed by the Minister of Education, Culture, Research and Technology in August 2023, where a thesis is no longer required and can be replaced with another form. The existence of this regulation has caused discussion in the community with various responses visible through social media Instagram. Instagram has become a forum for sharing the latest topics for the public, causing the emergence of online media accounts such as Folkative. This research aims to analyze sentiment regarding the policy of eliminating thesis obligations at universities through data collected from comments on Folkative's Instagram account posts. Sentiment analysis was carried out using Logistic Regression for classification, K-Fold Cross Validation for data validation, Confusion Matrix for evaluation, and ADASYN to resolve data imbalances. The data collected from the scraping process was 4,172 comments. After carrying out the classification and performance evaluation process, the best method produced was a combination of Logistic Regression and ADASYN achieving results of 89.6% accuracy, 91.5% precision, 87.7% recall, and f-1 score 89.5%. This shows that the performance of the Logistic Regression method is good and suitable for use as a classification model in this research.