Prediction of Dirty Cryptocurrency Using Machine Learning Models

Predicting the value of cryptocurrencies is essential due to their significant impact on the financial market and investment strategies. This paper evaluates the performance of three machine learning models: K-Nearest Neighbor (KNN), Linear Regression, and Random Forest Regression in predicting the value of various cryptocurrencies (Bitcoin, Ethereum, Bitcoin Cash, Litecoin, and Binance Coin (BNB)). Important metrics such as R-squared (R), accuracy, precision, recall, F1 score, Mean Squared Error (MSE), and Mean Absolute Error (MAE) are utilized alongside graphical comparisons of actual and predicted price categories. The analysis reveals that Random Forest Regression performs best for Bitcoin, while KNN is more suitable for Ethereum, BNB, and Litecoin. Linear Regression models tend to deviate more from actual prices. Both KNN and Random Forest Regression demonstrate a significantly closer relationship to real price categories. This study underscores the importance of model selection, highlighting the superior accuracy of KNN and random forest in predicting cryptocurrency prices. Future work could explore the integration of more advanced models and the impact of additional features on prediction accuracy.