Transaction Scam Detection Using Flask Frame Work

This research examines how to improve financial card default risk prediction by using Explainable AI (XAI) techniques, particularly the Locally Interpretable Model-agnostic Explanations (LIME) framework. This work uses LIME to shed light on the decision-making process of the machine learning models employed in credit card default prediction, in response to the growing demand for clear and understandable models in financial decision-making. This method uses a variety of classification methods, such as random forests (RF), support vector machines (SVM), and logistic regression (LR), to forecast default probabilities from a credit card dataset that is made publicly available. Performance measures, including Fl-score, recall, accuracy, and precision, are assessed to determine how effective the model is. Following that, LIME is utilized to describe both the global behavior of the models as well as individual forecasts. Key characteristics' impact on the predictions is demonstrated through visualizations such as force and prominence graphs. In addition to enhancing model interpretability, this study shows that integrating LIME guarantees an openness of economic danger assessments and gives stakeholders useful information about the significance of features. The findings show that debt management risk models that strike a balance between explain ability and model accuracy are more trustworthy and accountable, which helps users make better decisions in high -stakes banking applications.