Applying machine learning approach to exchange rate analysis

Abstract

The major concern of this study is to improve the forecasting accuracy of the exchange rate in time series analysis of foreign exchange stability. In the practical application, many traditional econometric forecasting models are difficult to apply to get satisfactory results with some being overly complex demanding high cost of computation, and some setting too many constraints or assumptions which cannot be realized in the real world. Although the techniques are not new, various deep learning or machine learning methods have motivated the present researcher to apply them to answer the study objective as the techniques have many recent successes in solving complex tasks to get the faster, less costly, and more reliable results. The problems addressed above have materialized motivations for this study.