ANALYSIS OF THE EFFECT OT TWITTER SENTIMENT ON BITCOIN PRICES USING THE SUPPORT VECTOR MACHINE AND MAXIMUM ENTROPY METHODS

• Main Author: Isyafira, Dyah
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/67347
• Institution: Institut Teknologi Bandung

Physics is a field of science that studies physical phenomena, and the development of physical science can be used to study human behavior. The diversity of human behavior in aspects of life such as social, cultural, and economic resembles the concept of complex systems in physics. In complex systems, interactions between the constituent parts of the system and macro and micro conditions will dynamically influence each other. One example of human behavior that resembles the concept of a complex system is public sentiment. The sentiment is the opinion of a group on a particular issue or topic. Twitter is a social media that is often used to express public opinion, for example, the case of cryptocurrency, which is currently being discussed by the world community, especially Bitcoin as the cryptocurrency with the largest market capitalization today. Departing from these conditions, the author is interested in seeing how Twitter's sentiment affects the Bitcoin price. Through the research results, negative emotion is very dominant, namely 51.96% of the total data used. The Support Vector Machine method produces 86.08% accuracy, 79.79% positive sentiment precision, 47.27% neutral sentiment precision, and 98.78% negative sentiment precision. Meanwhile, the Maximum Entropy method produces an accuracy value of 80.24%, a positive sentiment precision of 73.20%, a neutral sentiment precision of 22.37%, and a negative sentiment precision of 97.65%. Based on the results of these evaluation parameters, it can be seen that the Support Vector Machine method has a higher accuracy value than the Maximum Entropy method. However, the difference in the accuracy of these two methods is not too far, only 5.84%. Thus, it can be concluded that the Maximum Entropy method can be used for sentiment analysis.