Analisis Sentimen Pengguna Terhadap Ulasan Aplikasi Duolingo Menggunakan Metode Logistic Regression

Abstract

The ability to speak foreign languages, especially English, has become an important skill in the era of globalization and digitalization. However, according to the EF English Proficiency Index 2024, Indonesia ranks 80th out of 116 countries. One of the widely used solutions is the Duolingo application, a gamified language learning platform that has been downloaded over 500 million times. This research aims to analyze user sentiment toward the Duolingo application through reviews on Google Play Store using logistic regression. The data used consists of 8.648 reviews that have been labeled as positive and negative sentiment. The research process includes the stages of data preprocessing, dividing data into test and training data, weighting using TF-IDF, and classification using Logistic Regression algorithm with the parameter class_weight='balanced' to handle class imbalance, and evaluation using a confusion matrix. The evaluation results show that the model can achieve an accuracy of 89.83%, with a precision value of 73.91%, recall of 88.18%, and f1-score of 78.49%. This research shows that Logistic Regression with TF-IDF weighting is effective in sentiment analysis.