Design Implementation of Machine Learning Based Crypto Currency Prediction System

Abstract

The crypto currency market is a fast expanding marketplace for trade and investment that has attracted merchants, investors, and entrepreneurs on a global scale never seen before in this century. It will aid in documenting the behavior and habits of such a lucratively demanding and fast increasing sector by giving comparison studies and insights from the pricing data of crypto currency markets. In 2021, the bitcoin market is at one of its highest points ever. Crypto currencies have been more accessible to the general public as a result of the establishment of more exchanges, increasing their appeal. This, together with a number of credible crypto initiatives launched by some of the founders, has led to an increase in crypto currency users and interest. Virtual currencies are becoming increasingly popular, and companies such as Microsoft, Dell, and Tesla are now accepting them. As more people utilize decentralized digital currencies, it's more important than ever to appropriately inform the public about the new currencies as they grow in size, so that people know what they own and where their money is invested. We can answer our research question based on our findings: what price indicators may be used to predict the closing price of Bitcoin on a given day? The high-price, low-price, and open-price are all indicative of the same-day closing price, according to the model. Based on the coefficients discovered, we infer that the high-price has the biggest influence on the closing price. Contrary to popular assumption, the variable "volume" does not contribute to the information provided by the other variables about the closing price. We also observed that the "marketcap" parameter contains no further information. Because marketcap is normally computed using a product's pricing, this is unique. After deleting the two previously mentioned components, we found no significant change in adjusted R². It's worth noting that the model was created using data collected at the same period. The variables we gathered are made public at the same time as the closing price. We hope the model will be able to produce a decent prediction of what the closing price will be based on the current high and low prices for a given day.