Bitcoin Price Forecasting using LSTM and 10-Fold Cross validation

Abstract

This research paper reports the proposed model for price prediction of the popular Bitcoin crypto currency while applying different neural network approaches namely Recurrent Neural Network (RNN) and Long Short Term Memory (LSTM) along with 10-fold cross validation. In this work, the analysis of various trends of Bitcoin market is carried out and learning of important features used for price prediction is done. Daily price change is estimated by the neural network models. New activation functions are utilized in this research paper for improving efficiency. Further, this research paper compares the proposed model with other existing models namely; RNN with LSTM, Linear Regression and Random Forest applied in the same domain. The dataset utilized in this work is taken from the website named coinmarket and live streaming data is considered for the experimental work. Keras, Tensorflow and Scikit Learn have been used for performing the experimental work of the proposed model. The performance analysis of the proposed model with the existing ones has been carried out in terms of the Mean Absolute Error (MAE). It is observed from the results retrieved as a part of this work that the MAE for the proposed model came out to be 0.0043s which was significantly less than its existing counterparts.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 52 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.