Analisis Sentimen Ulasan Aplikasi Vidio pada Play Store Menggunakan Algoritma Support Vector Machine

Analisis SVM pada ulasan aplikasi vidio

Abstract

Vidio: Sports, Movies, Series is a popular streaming application in Indonesia with a 3.5 star rating on the Play Store in 2024. Although widely used, a number of user reviews reflect dissatisfaction with some of its features and performance. This research focuses on analyzing user opinions about the Vidio application on the Play Store and evaluating the accuracy of the SVM algorithm in grouping sentiments using the KDD method. KDD is applied to determine the analysis approach from the available data. In the data transformation stage, training and testing data are divided into three ratios: 90:10, 80:20, and 70:30. Words are weighted using the TF-IDF technique. This research tests SVM performance using four kernels: Linear, RBF, Poly, and Sigmoid. The research results show that a 70:30 ratio with the Sigmoid kernel produces the best performance, namely 83.33% accuracy, 86.13% precision, 83.33% recall, and 76.25% F1-score. This model succeeded in achieving an optimal composition between precision and recall, although there was a slight compromise seen in the F1- score.

Keywords— Sentiment Analysis, SVM, Vidio, Play Store