• Mushkan Tisekar
• Suja Jayachandran

In order to help investors and business owners make educated choices, precise price prediction algorithms have to be developed in light of the significant attention that cryptocurrencies have received in the financial industry. In this study, we combine the advantages of Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks to offer a novel approach for extracting essential characteristics from bitcoin price data. With the ability to recognize both short- and long-term relationships in datasets and samples, the LSTM-GRU architecture that has been implemented offers improved learning capabilities. Our model takes use of the sequential character of bitcoin pricing by fusing it with the strong memory retention of LSTM and GRU units. We enhance feature extraction using these recurrent neural network (RNN) versions, enabling the model to capture complex patterns and correlations across various monetary circumstances. A large dataset with a range of cryptocurrencies, time periods, and market situations is used to train the proposed LSTM-GRU fusion model. We use a thorough preprocessing method to standardize the input data, enabling consistent model performance across different currencies and time periods. We use an RNN-based prediction model that draws characteristics from an LSTM-GRU fusion layer to predict future bitcoin values. The aim of this forecasting model is to account for the inherent nonlinearity and volatility of bitcoin markets. Our methodology offers precise price projections that help traders and investors make wise choices by taking temporal dependencies and context into account. Using real-world bitcoin datasets, we evaluate the performance of our combined LSTM-GRU model against conventional LSTM and GRU architectures as well as other cutting-edge prediction models. By proving that our suggested strategy can accurately forecast cryptocurrency values across a range of currencies and time periods, the findings show how superior it is. In conclusion, our study offers a reliable and successful LSTM-GRU-based cryptocurrency price prediction model. We are able to strengthen feature extraction and prediction capabilities, leading to improved accuracy and generalization, by fusing the advantages of LSTM and GRU networks. The suggested model makes a substantial contribution to the developing area of cryptocurrency analytics and has the potential to greatly help investors and speculators in improving their investment strategies across various currency circumstances.

Cryptocurrency, Price Prediction, LSTM, GRU, RNN, Analysis

"Forecasting Price of Cryptocurrencies Using a LSTM-GRU Fusion Model", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.10, Issue 6, page no.h1-h7, June-2023, Available :http://www.jetir.org/papers/JETIR2306701.pdf

"Forecasting Price of Cryptocurrencies Using a LSTM-GRU Fusion Model", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.10, Issue 6, page no. pph1-h7, June-2023, Available at : http://www.jetir.org/papers/JETIR2306701.pdf