Default Prediction and Vehicle Recovery Probability Analysis in Auto Loans - A Segmentation Study

This study explains the segmentation model for loans under legal proceedings in a consumer finance company. Two machine learning models were developed: the first is a Loss Given Default (LGD) model that estimates the Recovery Rate (RR), and the second predicts the likelihood of locating pledged vehicles. Based on the outputs of these models, the files are divided into four segments using the K-means method and expert opinion. The dataset used in the study includes various financial, demographic, behavioral, credit variables, and macroeconomic indicators. The modeling process is conducted in light of existing methods in the literature.