Investigation of fluctuations in cryptocurrency transactions with sentiment analysis

This study aims to investigate the sentiment of the public towards popular cryptocurrencies listed on crypto exchanges in Turkey through comments shared on social media platforms and online forums. The high volatility and uncertainty of cryptocurrency markets pose significant challenges for investors in predicting market movements and minimizing risks. In this context, sentiment analysis plays a critical role in making accurate predictions and reducing risks. The study contributes to the literature by addressing the limitations of sentiment analysis studies conducted on Turkish texts. Data collected from social media platforms and online forums were analyzed using sentiment analysis techniques. A total of 607,592 comments were analyzed, classified as 89,986 negative, 72,655 positive, and 444,951 neutral. Binary classification was performed on 162,641 samples selected from 89,986 negative and 72,655 positive examples, training and testing machine learning models. The methodology of the study includes a detailed examination of sentiment analysis results obtained using various classifiers such as Naive Bayes, Logistic Regression, Decision Trees, K-Nearest Neighbors, Gradient Boosting, and Multi-Layer Perceptron. Findings illustrate how different cryptocurrencies are perceived across various social media platforms. For instance, BTC (Bitcoin) tends to be perceived more negatively on Investing.com and Telegram, while ETH (Ethereum) generally exhibits more negative sentiments. These results help investors understand perceptions and market expectations towards different cryptocurrencies. In conclusion, this study enhances understanding of the role of social media sentiment analysis in cryptocurrency markets, contributing to the development of new methods and approaches for future research.