AI-Driven Analysis of Credit Card Behaviour Score Using Machine Learning

Credit card behavior scoring is indispensable for banks which are eager to evaluate the credit worthiness of consumers, lower the risk, and adjust the lending strategy. The traditional methods are based on historical data and rule-based systems, which could overlook new types of consumer spending. In this paper, we have applied machine learning techniques to Kaggle credit card behavior score dataset. PCA is used in dimensionality reduction without loss of significant characteristics and to reduce redundancy. A k-means clustering is then carried out to classify customers according to their credit creation behavior and to identify three clusters of customers that represent low, medium and high users of risk. The findings of the research yield valuable information regarding customer segmentation and can assist financial institutions in developing or fine-tuning credit policies while allowing for early identification of a high-risk portfolio. It leads to more accurate and flexible credit scoring models and consequently to improved decision making under the financial risk assessment aspect.