default-prediction

Classification model to predict the probability that a customer defaults based on their monthly customer statements using the data provided by American Express.

NYU DS-GA 3001 Final Project

Logistic regression-based credit scoring model using public Kaggle data, designed for transparent PD estimation, performance evaluation, and teaching or regulatory use cases.

Machine learning model to identify customers that are more likely to default based on employment, bank balance and annual salary.

End-to-end Credit Risk Analytics project using Home Credit data featuring default prediction, XGBoost modeling, customer risk segmentation, underwriting framework, and Power BI dashboard.

Finance and Risk Analytics Project: Predicting credit default risk using machine learning models (Logistic Regression, Random Forest) and assessing stock market risk through historical returns and volatility analysis to guide financial risk management and investment strategies.

A program to take in loan level data and create a model which can predict probability of default

Machine learning project for credit card default prediction using CatBoost, probability calibration, SHAP explainability and cost-sensitive decision thresholds.

Working with an industrial scale data set to build a classification model to predict credit card default, and help creating a better customer experience for cardholders.

The goal of this project is to perform default prediction for commercial real estate property loans based on 17 variables.

Reproducible Python ML pipeline for credit-default scoring: src/, configs, validation, leakage checks, notebook report.

End-to-end credit risk modeling to predict loan default and support data-driven lending decisions.

Leakage-aware LendingClub default-risk prediction with logit, elastic net, CART, bagging, random forests, gains and lift screening, and cross-fitted DML.

Implementation of "Financial Default Prediction via Motif-Preserving Graph Neural Networks" - Demo application with synthetic financial network generation, structural pattern analysis, and GCN-based risk prediction.

Machine Learning pipeline to predict Long Overdue Debtors (LOD) on real-world financial loan data, optimizing ROC-AUC score and feature engineering workflows (Aihack Thailand 2025 Finalist).

AI-powered Loan Decision & Credit Risk Platform with Explainable AI, Risk Governance, Analytics Dashboard, and PDF Reporting built using Streamlit & Machine Learning.

Amex Default Prediction

Credit default prediction benchmark evaluated on KS, Gini, PSI and calibration — not accuracy. UNICAMP undergraduate research.

A group assignment on Machine Learning.

Production-ready credit risk modeling platform built with Streamlit and scikit-learn to predict loan default probability, generate 300–900 credit scores, explain decisions with SHAP, run what-if simulations, batch-score CSV files, and export PDF assessment reports.