Assessment of Support Vector Machine performance for default prediction and credit rating

  • 525 Views
  • 177 Downloads

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License

Predicting the creditworthiness of bank customers is a major concern for banking institutions, as modeling the probability of default is a key focus of the Basel regulations. Practitioners propose different default modeling techniques such as linear discriminant analysis, logistic regression, Bayesian approach, and artificial intelligence techniques. The performance of the default prediction is evaluated by the Receiver Operating Characteristic (ROC) curve using three types of kernels, namely, the polynomial kernel, the linear kernel and the Gaussian kernel. To justify the performance of the model, the study compares the prediction of default by the support vector with the logistic regression using data from a portfolio of particular bank customers. The results of this study showed that the model based on the Support Vector Machine approach with the Radial Basis Function kernel, performs better in prediction, compared to the logistic regression model, with a value of the ROC curve equal to 98%, against 71.7% for the logistic regression model. Also, this paper presents the conception of a support vector machine-based rating tool designed to classify bank customers and determine their probability of default. This probability has been computed empirically and represents the proportion of defaulting customers in each class.

view full abstract hide full abstract
    • Figure B1. Performance of the LR model (ROC)
    • Figure B2. ROC curve of the linear kernel
    • Figure B3. ROC curve of the Poly kernel
    • Figure B4. ROC curve of the Poly kernel
    • Table 1. Design of the scoring tool
    • Table 2. Portfolio allocation
    • Table 3. Confusion matrix
    • Table 4. Coefficients (w*i)
    • Table 5. Parameters of the selected kernels
    • Table 6. The three error ratios of the three kernels
    • Table 7. Portfolio allocation
    • Table 8. Conception of the rating tool
    • Table A1. Univariate analysis table
    • Table A2. The correlation table of the independent variables
    • Table A3. Wald test table
    • Table A4. Wald test table
    • Table A5. Test of the SVM-RBF parameters by the Gird-Search function
    • Table A6. Test of the SVM-Poly parameters by the function Gird-Search
    • Table A7. The confusion matrix of the three kernels
    • Table A8. Number of support vectors (RBF-SVM)
    • Conceptualization
      Karim Amzile, Mohamed Habachi
    • Data curation
      Karim Amzile, Mohamed Habachi
    • Formal Analysis
      Karim Amzile, Mohamed Habachi
    • Funding acquisition
      Karim Amzile
    • Investigation
      Karim Amzile, Mohamed Habachi
    • Methodology
      Karim Amzile, Mohamed Habachi
    • Project administration
      Karim Amzile, Mohamed Habachi
    • Resources
      Karim Amzile
    • Software
      Karim Amzile
    • Supervision
      Karim Amzile, Mohamed Habachi
    • Validation
      Karim Amzile, Mohamed Habachi
    • Visualization
      Karim Amzile, Mohamed Habachi
    • Writing – original draft
      Karim Amzile, Mohamed Habachi
    • Writing – review & editing
      Karim Amzile, Mohamed Habachi