Analisis Sentimen Terhadap Penggunaan Aplikasi Astro Menggunakan Algoritma Naïve Bayes

Abstract

Astro is a quick commerce application that focuses on providing fast groceries delivery services within less than an hour. It was established in 2021 under PT Astro Technologies Indonesia. Quick commerce applications are relatively new in Indonesia, and customer sentiments play a crucial role in determining the acceptance of this category within the society. Through reviews available on Google Play Store, we can gauge whether the majority of the populace holds a positive or negative sentiment towards it. The Astro app has been downloaded by over one million users, with a rating of 4.6 and more than 11.2 thousand reviews on Google Play Store. A total of 1,820 reviews were successfully obtained, and after preprocessing and data cleaning, 1,646 reviews remained. Subsequently, the data was labeled using the Naïve Bayes algorithm, categorizing sentiments into positive and negative. The results showed 1,213 reviews, or 74%, expressed a positive sentiment, while 433 reviews, or 26%, conveyed a negative sentiment. Using the confusion matrix, the classification performance achieved an accuracy of 63%, with a class recall of 61.58% for True Positive (TP) and 69.97% for True Negative (TN). The class precision for True Positive (TP) was 83.93%, and for True Negative (TN) was 38.36%. Based on the analysis, it can be concluded that the majority of Astro app users have a positive sentiment. However, a notable portion holds a negative sentiment, providing valuable insights to address concerns raised by users with negative sentiments.