Predictions of Cryptocurrency Prices Based on Inherent Interrelationships

Abstract

The price of cryptocurrencies is predicted in this paper based on their intrinsic interrelationship with Bitcoin. The Kaggle dataset is gathered, standardized, collated, and extracted. Convolutional Neural Network (CNN) is compared to other machine learning methods such as Linear Regression and K-Nearest Neighbor (KNN), and then parameter optimization is performed. The empirical results show that Linear Regression is less accurate than the other two models, whereas the CNN model employing end-to-end solutions outperforms other models with the best accuracy (overall above 0.95) forecasting the price quantitatively and directly of the majority of cryptocurrencies, despite the fact that forecasting takes a long time and tweaking its parameters is extremely time-consuming. This paper proposes using research object interrelationships rather than extrinsic relationships.