Yemeni Riyal Exchange Rate Prediction Using Predictive Models Based on Artificial Intelligent

Abstract

Foreign currency, such as the dollar, plays a fundamental role in controlling
market prices in many countries, because the import and export process takes
place through these foreign currencies. Predicting the exchange rates of these
changing currencies is a task with great challenges. The main purpose of this
paper is to build a system based on Artificial Intelligence to predict the
exchange rate between US Dollar (USD) and Yemen Riyal (YER) for the next
day or several days in the future. The artificial intelligence exploit two
techniques, Recurrent Neural Network (RNN) and Machine Learning (ML).
The proposed model uses recurrent Neural Network that exploits long short
term memory (LSTM) and machine learning with three models (Linear
Regression Model, Random Forest Regression and Gradient boosted
regression) to predict the currency USD/YER for the next day or several days
in the future with the highest accuracy. The objective function of training and
testing the prediction models to find out performance of proposed models by
calculating the root mean square error which came out to be very low. The
best results were obtained using Random Forest Regression model in root
mean square error training that reached 0.0448 and the best result in root mean
square error testing was obtained using Linear Regression model that reached
0.1070.