Cryptocurrency Price Prediction using Graph Embedding and Deep learning

Abstract

This paper displays the expectation of vacillation in the ensuing cost of the cryptocurrency. Foreseeing the cost of Bitcoin is an extraordinary test since it’s massively complicated and vital in nature. The major challenge for the investor is the volatile nature of cryptocurrency. We have attempted to foresee the coming cost of cryptographic forms of money like Bitcoin utilizing LSTM (Long Short-Term Memory), GRU (Gated Recurrent Unit) and LSTM-GRU-based Hybrid model. Throughout recent years, Blockchain innovation is becoming famous reasons are its different applications in various fields. One of the most widely recognized uses of Blockchain innovation is Cryptocurrency. Financial backers are vigorously putting resources into cryptocurrency. Yet, this cryptocurrency is impacted by an excessive number of variables. Here Price forecasts of digital money assume an essential part. It tends to be finished utilizing a few methods, however, AI and deep learning are the most popular procedures. AI is generally regularly utilized for forecasts however over the most recent couple of years; deep learning is additionally becoming famous due to its different applications in different fields. The cost of Cryptocurrency can anticipate by utilizing deep learning calculations. Our proposed technique is to anticipate the Cryptocurrency price with graph embedding and deep learning models. Neo4j sandbox is used for graph embedding. Different dimensions of graph embedding were taken to predict the price. The performance of the model was measured using RMSE, MSE and MAE.