In recent years, the digital currency has gained significant popularity owing to its increasing dependence on computers and the Internet. Among various forms of virtual currency, cryptocurrency has emerged as a prominent contender. The advent of digital currency has opened new avenues in the software industry, particularly in finance, data storage, and data collection. This evolution has given rise to exciting opportunities for businesses to explore the potential of digital currency and leverage its benefits. Cryptocurrency (crypto) is very volatile regarding the market value, which carries a host of unknowns that make it difficult to predict and analyze future prices. This paper discusses the use of six types of machine-learning models (Linear Regression, LSTM, Bi-LSTM, GRU, TARCH, and VAR) to predict the Bitcoin and DogeCoin prices; General Least-Squares Regression and Neural Networks algorithms to predict the volatility of a given cryptocurrency and its prices from 2014 to 2023 with daily cryptocurrency volatility data. The results show that high-performance computing techniques such as GRU (Gated Recurrent Unit) neural networks (0.0468 RMSPE) regression models to predict relatively accurate crypto price volatility and past available cryptocurrency price data are proven to be used to verify the prediction results.

近年来，由于数字货币对计算机和互联网的依赖程度越来越高，它获得了极大的普及。在各种形式的虚拟货币中，加密货币已成为突出的竞争者。数字货币的出现为软件行业开辟了新的途径，特别是在金融、数据存储和数据收集方面。这种演变为企业探索数字货币的潜力并利用其优势带来了令人兴奋的机会。加密货币（crypto）的市场价值波动很大，存在许多未知因素，导致难以预测和分析未来价格。本文讨论了使用六种机器学习模型（线性回归、LSTM、Bi-LSTM、GRU、TARCH 和 VAR）来预测比特币和狗狗币价格；通用最小二乘回归和神经网络算法，利用每日加密货币波动性数据来预测 2014 年至 2023 年给定加密货币的波动性及其价格。结果表明，GRU（门控循环单元）神经网络（0.0468 RMSPE）回归模型等高性能计算技术可以预测相对准确的加密货币价格波动性，并且过去可用的加密货币价格数据已被证明可用于验证预测结果。