Modelling of default risk for home credit data using machine learning approach
The banking industry plays an essential role in the financial system and economy of a nation. As a key component in the financial system, banks mainly function as the allocator of funds from savers to debtors. This major activity of borrowing and lending poses a large potential risk to the financial...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2022
|
| Subjects: | |
| Online Access: | http://eprints.sunway.edu.my/2395/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The banking industry plays an essential role in the financial system and economy of a nation. As a key component in the financial system, banks mainly function as the allocator of funds from savers to debtors. This major activity of borrowing and lending poses a large potential risk to the financial institutions as there is the prospect of default payment by the principal borrowers to the depositor bank. To mitigate such a possibility, financial risk management has been adopted by most financial institutions. In financial risk management, credit risk analysis holds substantial significance as the issuance of loans acts as the backbone for banks. The evaluation of customer credit which leads to the binary decision of loan acceptance or rejection is a lengthy yet tedious process which requires deep analysis of the quantitative and qualitative data fetched from the customer. However, with the debut and rise of financial technology, came a flood of newer modelling techniques such as machine learning. This study has as such surveyed and assessed three different modelling techniques that can be employed for credit risk analysis specifically for mortgage loan data classification. Logistic Regression (LR), Random Forest (RF) and Light Gradient Boosting Machine (LightGBM) were applied to a mortgage loan dataset obtained from the Home Credit Group corporation. As per the findings of this study, the anticipation of LightGBM’s predictive prowess over LR and RF for loan default were proven. This indicated the practicality of the LightGBM algorithm for the mortgage loan risk and prediction analysis of delinquent profiles for financial institutions. |
|---|