Delinquency Radar is a machine-learning model that detects early loan-delinquency risk using telecom micro-credit data. After data cleaning and feature engineering, a cost-sensitive XGBoost model handles imbalance. It supports MFS, banks, and micro finance by reducing defaults and improving financial inclusion
Jawaher AlQuraishi | DS-Sep/Dec-25
Lulwa alrushaid | DS-Sep/Dec-25
Maryam Alkanderi | DS-Sep/Dec-25
Rawan Mohsen Almutairi | DS-Sep/Dec-25
Shaikha Hussain Ali Al-Qahtani | DS-Sep/Dec-25
Programming Language: Python Data Processing: Pandas, NumPy Machine Learning: Scikit-learn, XGBoost Visualization: Matplotlib, Seaborn Model Evaluation: Accuracy, Precision, Recall, F1-score, Confusion Matrix Model: XGBoost Classifier
Costumer spending habits Period for loan repayment Average recharge amount
Users labeld 0 tend to pay back their loans on the same day Individuals who take out small loan amounts tend to require longer repayment durations Most customers fall within the medium-risk category Most customers are delinquent
Our AI model effectively predicts loan repayment behavior using telecom usage patterns, offering accurate and reliable risk insights. Feature analysis and smart recommendations support better, data-driven credit decisions. Looking ahead, we aim to enhance the model using longer historical time-series data to enable accurate forecasting and to support a more intelligent and automated decision-making system.