CRYPTOCURRENCY PRICE ANALYSIS WITH DEEP LEARNING (KEY IJP************260)

• Mrs. M.jayamma,G.renuka,M.kalyani,C.neelima,M.naga Sunitha,G.namitha

Abstract

Digital money has of late ignited far-reaching consideration among financial backers in light of the fact that to its hidden idea of decentralization and straightforwardness. Given digital money's instability and recognizing highlights, exact value expectation is basic for laying out powerful financial planning methodologies. To that reason, the creators of this paper present a special strategy for foreseeing the cost of Bitcoin (BTC), a main digital money. The change point identification approach is utilized to give consistent forecast execution in an unseen cost range. It is especially helpful for fragmenting time-series information so standardization might be performed separately founded on division. Besides, on-chain information, or exceptional records posted on the blockchain that are characteristic in digital currencies, is accumulated and utilized as info factors to expect values. Also, for the expectation model, this paper presents self-consideration based numerous long short-term memory (SAM-LSTM), which includes different LSTM modules for on-chain variable gatherings and the consideration instrument. Tests utilizing genuine BTC cost information and different procedure settings have shown that the proposed structure is viable at anticipating BTC costs. Results are positive, with the best MAE, RMSE, MSE, and MAPE upsides of 0.3462, 0.5035, 0.2536, and 1.3251, individually.