Publication

Abstract : Cryptocurrencies have become a major element in enterprises and financial market showing promising potential during the past ten years. Predictions that are accurate can help cryptocurrencies investors to make the best decisions and possibly enhance their earnings. For this work 4 cryptocurrencies namely Bitcoin, Binance Coin, Ethereum and Tether (USDT) are considered. The considered dataset is for around 5 years, which has a total of 9 columns, One Date Column ranging from year 2017 to 2022 and for each of the Coin there is the Closing price and the Volume. The Dataset is collected from Kaggle. Attempts are made to build the models in various ways, by dealing with certain features or by taking a subset of the dataset. All the models, according to the way the dataset was dealt were evaluated based on certain evaluation metrics like R-squared, mean square error and root mean square error. The models used are KNN, Decision Tree, Random Forest, XGBoost and CatBoost. Visualized and understood how the prices of those cryptocurrencies were impacted by other features of the dataset. The models that outperformed are Random Forest and CatBoost. Achieved RMSE was 365, 2.99, 29.99 and 0.0023 for Bitcoin, Binance, Ethereum and USDT respectively.