DeepFinancial Model for Exchange Rate Impacts Prediction of Political and Financial Statements

Abstract

The extensive use of social media led people to share emotions and opinions on social sites. Currently, the prediction of the exchange rate with the content of social sites, specifically Twitter, is an active research and challenge. In this study, we have proposed a deep learning model for the prediction of the exchange rate fluctuation with political and financial statements sentiments. In this study, we have considered USD dollar rates in terms of PKR currency rates for experiments as well as collective sentiment technique (positive, negative, and neutral for each day) considered after data preprocessing with natural language processing techniques. The Adaptive Synthetic (ADASYN) technique is used in this study for data balancing to avoid the overfitting of the machine and deep learning models. Deep learning based proposed model named “Deep Financial” is effective with the highest accuracy of 87.54% as compared to Support Vector Machine, K-Nearest Neighbor and Logistic Regression, for the prediction of exchange rate fluctuation with political and financial statements sentiments.