Trend analyses and price predictions on cryptocurrencies by using artificial intelligence

Since the introduction of Bitcoin (BTC) for the first time in 2008 by Satoshi Nakamoto through his white paper, concept of cryptocurrency or block-chain currency where encrypted data string is used in an architecture of decentralized network and transparent ledger has become the basis of today's digital currencies. Today there are more than 25,000 cryptocurrencies like BTC, Ethereum (ETH), Litecoin (LTC), Ripple (XRP), Monero (XMR), etc. and prediction of their future values is a major task. In this work, trend analyses and price predictions of cryptocurrencies by using artificial intelligence (AI) is being studied. AI is a wide-ranging interdisciplinary branch of computer science concerned with building smart machines capable of performing tasks that normally require human intelligence. It involves the ability of machines to replicate or enhance human intelligence, such as reasoning and learning from experience. AI has a main subclass of Machine Learning (ML) which has another subclass called Deep Learning (DL). From literature it seems that, beside statistical methods, some certain methods like Artificial Neural Network (ANN) and Linear Regression (LR) in the ML and Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) in the DL are used to predict future price values of crypto-currencies. In this work, we study two fundamental AI regression techniques, namely, LR and ANN in two different prediction models to predict the future value of the most popular crypto-currency, BTC, by using different candle-stick datasets obtained from Yahoo-finance websites via Python codes. Next ten upcoming BTC values outside the five different datasets studied here are predicted successfully.