An Efficient Deep Learning Model for Financial Data Analysis: A Comparative Evaluation with Traditional Machine Learning

Abstract

Due to the rapidly increasing amount of financial data, there is an increasing demand for the development of reliable, fast algorithms that could detect complicated patterns and help in making financial decisions. In this paper, a deep learning method is proposed for the analysis and enhancement of financial data through a Deep Neural Network model, and its performance is compared with Gradient Boosting. The LAR dataset was chosen for testing and training the two algorithms in the same experimental setting. The DNN architecture used in this paper was devised in a way that it could automatically detect and learn from complicated patterns present in financial data. The experimental results reveal that the (Efficient-DNN) model obtained 94.35% ROC-AUC, while Gradient Boosting obtained 91.30%.

The results revealed that the (Efficient-DNN) model can be considered an efficient tool for obtaining higher predictive accuracy in analyzing financial data due to its ability to learn high-level features from the input data. The experimental results confirmed the applicability of deep learning methods for financial an alhakam.a.allawie@tu.edu.iq   alytics and proved that the proposed model is a good alternative to traditional ensemble learning models.