This paper presents machine learning-based credit card approval prediction and gives an overview of the machine learning models and algorithms that are used to authorize credit cards for users. In order to improve credit card acceptance predictions and increase accuracy and adaptability in financial risk assessment, this study uses XGBoost in machine learning. The study emphasizes the importance of XGBoost in addressing challenges such as handling missing data, avoiding overfitting, and efficiently managing large datasets. Comparisons between the decision tree classifier and XGBoost reveal the latter's advantages, including interpretability, ability to handle complex relationships, and efficiency in processing large datasets. Results from experiments using the XGBoost algorithm demonstrate an accuracy of 90.06%, affirming its efficacy in credit card approval prediction.