|Title||Interpretable machine learning credit scoring model.|
|Authors/Creators||Foo Yeong Jin (TP044538)|
This study divided into two distinct part to accurately assess the interpretability of the proposed model. The first part is to develop a credit scoring model using the black-b ox algorithm, and the second part is to develop an interpretable model to interpret the black-box model. The extreme gradient boosting (xgboost) is used as the black-box algorithm and Local Interpretable Model- agnostic Explanations (LIME) is used as the interpretation model. For the evaluation, this study is using two different assessment to measure the interpretability of the interpretation model.
The first assessment is to measure the consistency through the weight assigned to each independent variable in both training and validation data. The second assessment is the interpretability, which the signage assigned to each independent variable is used to compare against the finding from the literature. The result shows that the proposed model is achieved the desired consistency and interpretability. The difference between the training and validation data is less than 5%. The interpretability is also same as the finding gathered from the literature.
|Supervisor||V. Sivakumar, Dr.|
|Institution||Asia Pacific University of Technology and Innovation (APU)|
|School||Graduate School of Technology|
|No. of pages||63|
|Refereed||Yes, this version has been refereed|
A thesis submitted in fulfillment of the requirement of Asia Pacific University of Technology and Innovation for the award of the degree of MSc in Data Science and Business Analytics (UCMP1701DSBA).
Credit scoring systems ; Machine learning ; Regression analysis ; Data processing.
|View||Login to view full text|