Modelling and Forecasting the Short-term Bitcoin Prices using Bayesian Neural Networks
A research proposal practice based on [12]H. Jang and L. Jaewook’s paper.
Abstract
Since its invention, Bitcoin has gained amazing popularity and much attention in various research fields, including computer science, economics and cryptography. With the emergence of the Blockchain technology, the innovative ledger technology underpinning Bitcoin. This study employs Bayesian neural networks (BNN) to model and predict the short-term Bitcoin prices. Some most relevant factors like Blockchain information, macroeconomic factors and foreign exchange rates are selected as input features to improve the forecasting accuracy of proposed model. A comparative analysis is also planned to evaluate the prediction performance between BNN and other two linear and non-linear benchmark methods. Finally, the performance of time series forecasting on Bitcoin prices will be validated with trained BNN model.
