Previsões de preço de criptomoedas utilizando algoritmos de Aprendizado de Máquina

Abstract

Cryptourrencies are decentralized digital currencies created with the intent of replacing traditional forms of payment. Their high volatility is one of the aspects that causes distrust among investors. Artificial Intelligence based on Machine Learning algorithms can be an invaluable tool to analyse this characteristic, helping to predict prices. In this sense, this project’s goal is to use algorithms implemented in Python — Support Vector Machine and Long Short-Term memory — alongside convenient performance metrics such as mean squared error, mean absolute error, maximum error and R^2 to assess their price prediction capabilities in face of such randomness.