Because of the high instability and unpredictability of Bitcoin prices, it is important to develop robust machine learning models to predict price fluctuations accurately. This paper explores the prediction of Bitcoin price using five high-performance regression models. To test the predictive performance of the models, a benchmark dataset including the Bitcoin price from the past 10 years, with 22 variables like trading volume and technical indicators, is used. Experimental results indicate that the orthogonal matching pursuit (OMP) outperformed other models regarding performance metrics. The OMP obtained MAE=297.06, MSE=251046.3, and $R^2$=0.99 in the testing dataset.
