Analisis Sentimen Pengguna Instagram terhadap Timnas Indonesia U-23 pada Piala AFC menggunakan Algoritma K-Nearest Neighbor (K-NN) dengan SMOTE

Abstract

The Indonesia U-23 National Team is Indonesia's national football team with players under the age of 23. The Indonesian public's sentiment towards the Indonesia U-23 National Team increased rapidly during the 2024 AFC Cup period. This trend can be observed on one of the social media platforms, Instagram. Instagram is widely used by the public today. By using Instagram, we can obtain various types of information shared by the public. Within this information, there is data that can be processed into sentiment analysis. Based on this premise, a study was conducted using the K-Nearest Neighbor (k-NN) method to analyze public sentiment towards user comments on Instagram. Data was collected using scraping techniques. After that, the data was cleansed and preprocessed. Then, the data was labeled and extracted using TF-IDF before being classified with K-Nearest Neighbor (k-NN). The dataset obtained from the scraping technique was imbalanced, which can affect the model's accuracy. Therefore, in this study will compared the performance of k-NN with SMOTE and k-NN without SMOTE. The experiments showed that the imbalanced data achieved only 50% accuracy for k = 3. After oversampling using SMOTE, the accuracy increased to 55% for k = 3.