Credit Risk Modeling in Python

Curriculum

• Introduction Free

  6 Lessons 38 Min

  We start by explaining why credit risk is important for financial institutions. We also define ground 0 terms, such as expected loss, probability of default, loss given default and exposure at default.

  What does the course cover Free What is credit risk and why is it important? Free Expected loss (EL) and its components: PD, LGD and EAD Free Capital adequacy, regulations, and the Basel II accord Free Basel II approaches: SA, F-IRB, and A-IRB Free Different facility types (asset classes) and credit risk modeling approaches Free
• Setting up the environment Free

  2 Lessons 1 Min

  Here you will learn how to set up Python 3 and load up Jupyter. We’ll also show you what the Anaconda Prompt is and how you can use it to download and import new modules.

  Setting up the environment Free Installing the relevant packages Free
• Dataset description Free

  2 Lessons 9 Min

  Our example focuses on consumer loans. Since there are more than 100 potential features, we've devoted a complete section to explain why some features are chosen over others.

  Our example: consumer loans. A first look at the dataset Free Dependent variables and independent variables Free
• General preprocessing Free

  4 Lessons 27 Min

  Each raw datasets has its drawbacks. While most preprocessing is model specific, in some cases (like missing values imputation), we could generalize the data preparation.

  Importing the data into Python Free Preprocessing few continuous variables Free Preprocessing few discrete variables Free Check for missing values and clean Free
• PD model: data preparation

  15 Lessons 113 Min

  Once we have completed all general preprocessing, we dive into model-specific preprocessing. We employ fine classing, coarse classing, weight of evidence and information value criterion to achieve the probability of default preprocessing. Conventionally, we should turn all variables into dummy indicators prior to modeling.

  How is the PD model going to look like? Dependent variable: Good/ Bad (default) definition Constructing independent variables Information value Data preparation. Splitting data Data preparation. Preprocessing one discrete variable Data preparation. Preprocessing discrete variables: automating calculations Data preparation. Preprocessing discrete variables: visualizing results Data Preparation. Preprocessing Discrete Variables: Creating Dummies (part 1) Data Preparation. Preprocessing Discrete Variables: Creating Dummies (part 2) Data preparation. Preprocessing continuous variables: automating calculations Data preparation. Preprocessing continuous variables: creating dummies (part 1) Data preparation. Preprocessing continuous variables: creating dummies (part 2) Data preparation. Preprocessing continuous variables: creating dummies (part 3) Data preparation. Preprocessing the test dataset
• PD model estimation

  5 Lessons 35 Min

  Having set up all variables to be dummies, we estimate the probability of default. The most intuitive and widely accepted approach is to employ a logistic regression.

  The PD model. Logistic regression with dummy variables Loading the data and selecting the features PD model estimation Build a logistic regression model with p-values. Interpreting the coefficients in the PD model
• PD model validation (test)

  3 Lessons 28 Min

  Since each model overfits the training data, it is crucial to test the results on out-of-sample observations. Consequently, we find its accuracy, its area under the curve (AUC), the Gini coefficient and the Kolmogorov-Smirnov test.

  Out-of-sample validation (test). Evaluation of model performance: accuracy and area under the curve (AUC) Evaluation of model performance: Gini and Kolmogorov-Smirnov.
• Applying the PD model for decision making

  5 Lessons 36 Min

  In practice, banks don't really want a complicated Python-implemented model. Instead, they prefer a simple score-card which contains only yes and no questions that could be employed by any bank employee. In this section, we learn how to create one.

  Calculating probability of default for a single customer Creating a scorecard Calculating credit score From credit score to PD Setting cut-offs
• PD model monitoring

  3 Lessons 28 Min

  Model estimation is extremely important, but an often-neglected step is model maintenance. A common approach is to monitor the population stability over time using the population stability index (PSI) and revisit our model if needed.

  PD model monitoring via assessing population stability Population stability index: preprocessing Population stability index: calculation and interpretation
• LGD and EAD models

  3 Lessons 17 Min

  To calculate the final expected loss, we need three ingredients: probability of default (PD), loss given default (LGD) and exposure at default (EAD). In this section, we preprocess our data to be able to estimate the LGD and EAD models.

  LGD and EAD models: independent variables LGD and EAD models: dependent variables LGD and EAD models: distribution of recovery rates and credit conversion factors
• LGD model

  7 Lessons 27 Min

  LGD models are often estimated using a beta regression. To keep the modeling part simpler, we employ a two-step regression model, which aims to simulate a beta regression. We combine the predictions from a logistic regression with those from a linear regression to estimate the loss given default.

  LGD model: preparing the inputs LGD model: testing the model LGD model: estimating the accuracy of the model LGD model: saving the model LGD model: stage 2 – linear regression LGD model: stage 2 – linear regression evaluation LGD model: combining stage 1 and stage 2
• EAD model

  2 Lessons 10 Min

  The exposure at default (EAD) modeling is very similar to the LGD one. In this section, we take advantage of a linear regression to calculate EAD.

  EAD model estimation and interpretation EAD model validation
• Calculating expected loss

  1 Lesson 16 Min

  After having calculated PD, LGD, and EAD, we reach the final step: computing expected loss (EL). This is also the number which is most interesting to C-level executives and is the finale of the credit risk modeling process.

  Calculating expected loss