In the dynamic global economy, for any future investment, the better accuracy and correctness of Foreign exchange value is important. The Foreign exchange market is facing unprecedented growth over the last few years. Hence the accurate prediction of foreign exchange is very crucial for global investing and international businesses. It is very important to predict the forex rate, as it will help investors make better decisions to minimize risks and gain more returns. In modern time series forecasting, the foreign exchange prediction is one of the important and demanding applications. The forex rates are noisy, chaotic and non-stationary. This characteristic shows there is no clear past behavior employs in the data and there is no connection that we can explain clearly from the future data from that of the past.One assumption that we can make is that the historic data have all the behaviors incorporated in it. Hence the historic data helps to predict the future values better. However we cannot say how good this prediction is.
In this project we plan to design and implement ARIMA time series model that can predict future rates and its trends. We also implemented ARIMA with XGBoost methodology using the same dataset to evaluate which model is more accurate.