Analisis Sentimen Ulasan Pengguna Aplikasi Crunchyroll, iQIYI, Wibuku di Google Play Store Menggunakan Metode Random Forest

Fathon, Ulil
Arifin, Muhammad
Setiawan, Arif

Jurnal Sistem Informasi Dan E-Bisnis
Vol 7 No 2 (2025): Agustus

This study evaluates the performance of the Random Forest model in classifying sentiment in reviews on three platforms, namely Crunchyroll, iQIYI, and Wibuku. The evaluation results show an accuracy of 82% on Crunchyroll, 83% on iQIYI, and 84% on Wibuku. On Crunchyroll, the model demonstrated superiority in recognizing negative reviews (precision 88%, recall 84%), while on iQIYI and Wibuku, the best performance was achieved on positive sentiment with a precision of 94% and 92%, respectively. However, the precision values for negative sentiment on iQIYI (62%) and Wibuku (66%) indicate room for improvement, particularly in enhancing the model's ability to identify negative-toned reviews. Overall, Random Forest proved effective for cross-platform sentiment analysis; however, further optimization is needed to improve the performance balance across sentiment categories.