orestismk / orestismk-multivariate-forecast-with-var-svr-rnn-lstm

Forecasting exchange rates by using commodities prices

OrestisMk-Multivariate-forecast-with-VAR-SVR-RNN-LSTM

Those forecasting models was part of my thesis, with name "FORECASTING EXCHANGE RATES BY USING COMMODITY PRICES WITH TRADITIONAL ECONOMETRIC, MACHINE LEARNING AND DEEP LEARNING TECHNIQUES", at City University of London.

Abstract :

The accuracy of the prediction of economic parameters and the correct estimation of foreign exchange rates are of paramount importance in modern global economy. For this reason, the use of computational techniques, which are based on technology, in order to forecast macroeconomic variables is essential and highly recommended. This project intends to make a multistep forecast by using multivariate times series in order to predict the exchange rates of the Australian dollar while using the most tradeable exports of Australia. To do so, the Vector Auto Regression model, the Support Vector Machine model and the Recurrent Neural Networks with Long Short-Term Memory model have been used. Also, monthly historical data for five macroeconomic variables (Gold, Iron ore, Natural gas, Crude oil (brent), Coal) have been used for the period between August 1995 and January 2020 to predict the Australian Foreign Exchange Rate.