Applications of Explainable AI in Financial Time Series

Deep learning models have evolved significantly, achieving high performance in forecasting. However, they are often referred to as black-box models because their decision-making process remains unclear. Furthermore, the AI Act in Europe requires AI models in high-risk categories to be more transparent. In this research, we applied TimeSHAP to a financial use case and investigated its open issues and research directions. Beyond explaining the machine learning model, this study also explores how to make the entire machine learning pipeline more explainable by incorporating ontologies based on the Explainable ML Pipeline Ontology (XMLPO)