Currency Exchange Rate Forecasting using Associative Models

Associative Models were created and used for pattern recognition tasks, but recently such models have shown good forecasting capabilities; by a preprocessing of a time series and some fit of the Model. In this paper, the Gamma Classifier is used as a novel alternative for currency exchange rate forecasting, where experimental results indicate that the proposed method can be effective in the Exchange Rate Time Series Prediction, compared to classical Machine Learning Models (ANN, SVM, MLP) and well known for the Financial and Economy Fields Box-Jenkins Models (AR, ARMA, ARIMA).