Forecasting real exchange rate (REER) using artificial intelligence and time series models.

Area of Science:

• Economics
• Data Science
• Financial Modeling

Background:

• Accurate forecasting of economic phenomena is crucial for market understanding.
• Machine learning (ML) algorithms offer advanced capabilities for analyzing complex data patterns.
• Real Exchange Rate (REER) data is a significant indicator in the business and financial markets.

Purpose of the Study:

• To model and predict Real Exchange Rate (REER) data using various machine learning and time series models.
• To identify the most effective forecasting model for REER based on performance criteria.
• To contribute to the understanding of exchange rate dynamics through advanced analytical techniques.

Main Methods:

• Utilized machine learning models: Multi-layer perceptron (MLP) and Extreme learning machine (ELM).
• Employed classical time series models: Autoregressive integrated moving average (ARIMA) and Exponential Smoothing (ES).
• Applied models to REER data from January 2019 to June 2022 (864 observations), splitting into training and testing sets.

Main Results:

• All models were applied and evaluated against Key Performance Indicators (KPI).
• A comparative analysis was performed to determine model efficacy.
• The study identified a best-performing candidate model for REER prediction.

Conclusions:

• The selection of an appropriate forecasting model is critical for accurate economic predictions.
• Machine learning models demonstrate potential in capturing complex patterns in financial time series data.
• The chosen model provides a robust approach for predicting future real exchange rate behavior.