Machine Learning-Based Forecasting Model for Bitcoin Price Prediction

Abstract

The decentralized structure, lack of regulation, and susceptibility to manipulation of Bitcoin markets result in a high level of volatility, which presents substantial obstacles to the accurate prediction of prices. Traditional statistical models, like ARIMA, frequently fall short in describing the dynamic and nonlinear nature of bitcoin markets. In order to overcome this constraint, this research utilizes sophisticated machine learning and deep learning techniques, including as Convolutional Neural Networks (CNN), Decision Trees, Long Short-Term Memory (LSTM), and Logistic Regression, to predict changes in the price of Bitcoin. Using historical Bitcoin datasets, the suggested models are trained and assessed using performance measures like accuracy and RMSE. In comparison to traditional techniques, experimental results show that deep learning models—in particular, LSTM—achieve greater prediction accuracy, offering a more dependable framework for forecasting bitcoin prices.