A comparison of default prediction models trained on different datasets: model-based clustering

Yaozhi Yang

doi:10.1117/12.2627614

11 March 2022 A comparison of default prediction models trained on different datasets: model-based clustering

Yaozhi Yang

Proceedings Volume 12160, International Conference on Computational Modeling, Simulation, and Data Analysis (CMSDA 2021); 121600I (2022) https://doi.org/10.1117/12.2627614
Event: International Conference on Computational Modeling, Simulation, and Data Analysis (CMSDA 2021), 2021, Sanya, China

Abstract

This paper compares the models trained on the different mortgage markets of different states in the USA. We use the measured performance (AUC) to generate a similarity matrix and then generate clusters. We regard AUC as a similarity between two models and use t-SNE method combined with k-means to construct clusters and visualize them. There are two main findings. One is that the model of MA (Massachusetts) is relatively similar to the model of CA (California) not NY (New York) although MA is next to NY and far from CA. The other is that the crisis in 2008 makes the differences between the model of OK (Oklahoma) and the model of TX (Texas) more obvious.

Citation Download Citation

Yaozhi Yang "A comparison of default prediction models trained on different datasets: model-based clustering", Proc. SPIE 12160, International Conference on Computational Modeling, Simulation, and Data Analysis (CMSDA 2021), 121600I (11 March 2022); https://doi.org/10.1117/12.2627614

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available