The IUP Journal of Computer Sciences
Multilayer Perceptron Model for Macroeconomic Forecasting of Cryptocurrency Market in Nigeria

Article Details
Pub. Date : Oct, 2022
Product Name : The IUP Journal of Computer Sciences
Product Type : Article
Product Code : IJCS031022
Author Name : Eucharia Onyekwere, Francisca Nonyelum Ogwueleka and Martins Ekata Irhebhude
Availability : YES
Subject/Domain : Management
Download Format : PDF Format
No. of Pages : 20

Price

Download
Abstract

Financial time-series data derived from the volume of cryptocurrency transactions is complicated and challenging to predict due to the volatility and uncertainty of the digital market. In this paper, the Multilayer Perceptron (MLP) models with Backpropagation Algorithm (BPA) were designed to forecast the volatile time-series data of the volume of Bitcoin transactions in Nigeria. Three different models were designed, with one hidden layer, two hidden layers and three hidden layers. Evaluation of the models was based on performance measures, and the results show that model 2 with two hidden layers outperformed the other two models. Furthermore, the Autoregressive Integrated Moving Average (ARIMA) model was used as a benchmark for the MLP models. Compared to the ARIMA model, the MLP models had a percentage accuracy of 98.73%, 98.81% and 97.2% for model 1, model 2 and model 3, respectively. In contrast, the ARIMA model had a percentage accuracy of 53.87%.


Introduction

Cryptocurrency is a Peer-to-Peer (P2P) digital currency system that electronically permits payment from one individual or entity to another, skipping all forms of financial institutions. Unlike fiat currency, it is markedly divisible, virtual and has no affiliation with a higher authority. In contrast to conventional financial assets, its value is neither based on a tangible asset, an organization, or a country's economy; instead, it is based on the security of an algorithm that can trace all transactions. Corbet et al. (2019) and Kurka (2019) observed that cryptocurrency had attracted investors, media and regulators in the financial industry since the Bitcoin was introduced in 2009. Out of all


Keywords

Autoregressive Integrated Moving Average (ARIMA), Artificial Neural Network (ANN), Bitcoin, Cryptocurrency, Multilayer Perceptron (MLP)