Software Design Challenges in Time Series Prediction Systems Using Parallel Implementation of Artificial Neural...

Area of Science:

• Computer Science
• Data Science
• Financial Modeling

Background:

• Traditional software development faces challenges with predictive modeling, especially for time series data like currency exchange rates.
• Statistical models were initially used for financial forecasting, but Artificial Neural Networks (ANNs) have gained prominence in the last two decades for improved accuracy.
• Existing methods often struggle with integrating diverse data strengths for robust predictions.

Purpose of the Study:

• To address architectural design issues for performance enhancement in predictive software development.
• To propose an adaptive, hybrid methodology for predicting exchange rates.
• To evaluate the accuracy and performance of parallel algorithms within this new framework.

Main Methods:

• Vectorizing the strengths of multivariate econometric time series models.
• Integrating Artificial Neural Networks (ANNs) for enhanced predictive capabilities.
• Developing and testing a hybrid framework for exchange rate forecasting using parallel algorithms.

Main Results:

• The proposed hybrid methodology demonstrates improved performance in predicting exchange rates.
• Architectural design improvements lead to enhanced accuracy in time series forecasting.
• The framework effectively leverages parallel algorithms for computational efficiency.

Conclusions:

• The hybrid approach offers a significant advancement over traditional methods for financial time series prediction.
• This research provides a robust framework for accurate and efficient exchange rate forecasting.
• The integration of econometric models and ANNs presents a promising direction for predictive analytics.