2026

Journal

Computational Economics, 1-62.

This paper introduces TunPredML, an intelligent financial decision support system designed to predict stock market prices under volatile and uncertain conditions. The system integrates machine learning models, advanced data analytics, and visualization techniques to enhance financial forecasting and decision-making for investors. First, we analyze the Tunisian stock market’s reaction to recent crises, including financial, political, and health disruptions, to assess their impact on market volatility and losses. Next, we conduct an initial experiment using eight machine learning algorithms: Linear Regression (LR), Support Vector Regression (SVR), eXtreme Gradient Boosting (XGBoost), Simple Recurrent Neural Network (SRNN), Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), Artificial Neural Network (ANN), and Random Forest (RF). These models are evaluated on the Tunindex, incorporating event-related data, and assessed using four regression metrics in addition to execution time to ensure both predictive accuracy and computational efficiency. Based on the evaluation results, the most accurate models are selected for a second experiment, where they are combined into an ensemble stacked model to improve predictive performance. To enhance accessibility and usability, TunPredML is deployed as a web-based financial decision support system, offering interactive data visualization and real-time analytics. The platform provides investors and traders with key insights, including model performance metrics, graphical comparisons of predicted versus actual stock prices, error distributions, and 10-day future price forecasts. By leveraging cutting-edge machine learning techniques and a user-friendly digital platform, the proposed system empowers investors to mitigate risks, navigate market uncertainties, and make well-informed investment decisions. The effectiveness of TunPredML is validated using real-world financial data, demonstrating its potential as a robust, data-driven forecasting tool for financial markets, particularly during periods of economic instability.

@article{zouaghia2026tunpredml,
  title={TunPredML: A Machine Learning-Based Financial Decision Support System for Crisis-Aware Stock Market Forecasting and Risk Mitigation: Empirical Insights from the Tunisian Stock Market},
  author={Zouaghia, Zakia and Kodia, Zahra and Ben Said, Lamjed},
  journal={Computational Economics},
  pages={1--62},
  year={2026},
  publisher={Springer}
}