Session

LAW

Description

Mortgage scoring models are pivotal in evaluating the risk associated with mortgages. Traditionally, these models were constructed using logistic regression. However, with the rise of machine learning, algorithms such as classification trees and neural networks have been employed. These algorithms are trained on a sample of mortgages, with the occurrence or non-occurrence of default observed. The data is then split into training and test samples, with machine learning algorithms further dividing the training sample for validation. This approach aims to determine hyperparameters that maximize performance while minimizing overfitting. Once calibrated, the model is applied to the test sample to predict default events. Despite the sophistication of machine learning algorithms, their predictive performance in mortgage scoring is comparable to logistic regression. Ensemble methods, which combine multiple models, have shown potential in enhancing predictive performance. This literature review explores the application of machine learning in mortgage scoring, comparing it with traditional methods, and discussing its implications.

Keywords:

scoring model, mortgage default, machine learning algorithms, logistic regression, receiver operating characteristic curve, neural networks

Proceedings Editor

Edmond Hajrizi

ISBN

978-9951-550-95-6

Location

UBT Lipjan, Kosovo

Start Date

28-10-2023 8:00 AM

End Date

29-10-2023 6:00 PM

DOI

10.33107/ubt-ic.2023.124

Included in

Law Commons

Share

COinS
 
Oct 28th, 8:00 AM Oct 29th, 6:00 PM

Enhancing Real Estate Management: The Transformative Role of Machine Learning in Predictive Gains and Risk Model Performance

UBT Lipjan, Kosovo

Mortgage scoring models are pivotal in evaluating the risk associated with mortgages. Traditionally, these models were constructed using logistic regression. However, with the rise of machine learning, algorithms such as classification trees and neural networks have been employed. These algorithms are trained on a sample of mortgages, with the occurrence or non-occurrence of default observed. The data is then split into training and test samples, with machine learning algorithms further dividing the training sample for validation. This approach aims to determine hyperparameters that maximize performance while minimizing overfitting. Once calibrated, the model is applied to the test sample to predict default events. Despite the sophistication of machine learning algorithms, their predictive performance in mortgage scoring is comparable to logistic regression. Ensemble methods, which combine multiple models, have shown potential in enhancing predictive performance. This literature review explores the application of machine learning in mortgage scoring, comparing it with traditional methods, and discussing its implications.