<![CDATA[The advancement of digital technology has driven a significant transformation in the investment landscape, particularly through digital assets such as cryptocurrency. Tokocrypto, as one of the local crypto exchanges in Indonesia, has become the focus of social media users, especially on X (formerly Twitter), who often express their opinions about the platform. To understand user sentiment, a sentiment analysis was conducted using 1,386 tweets collected with the keywords “investasi tokocrypto” and “cex tokocrypto”, which were then manually filtered to remove irrelevant data, resulting in a final dataset of 600 tweets. The data was collected through a crawling method and cleaned through preprocessing steps including case folding, normalization, tokenizing, stopword removal, and stemming. After labeling, neutral sentiments were removed, leaving 417 tweets categorized into positive and negative sentiments. The classification process was carried out using the K-Nearest Neighbor (K-NN) method, evaluated across various training and testing ratios and different K values. The best-performing model was achieved with K=7 and an 80:20 ratio, producing the highest performance with accuracy = 74%, precision = 75%, F1-score = 78%, and recall = 81%. The results of this study demonstrate the effectiveness of the K-Nearest Neighbor method in sentiment analysis and provide insights into user perceptions of Tokocrypto as an investment platform on social media platform X.]]>
