Module 0: Quantum Computing and Algorithms

• Exercise 1: Quantum operations

Module 1: Artificial Intelligence for Credit Scoring

(No exercises)

Module 2: AI in Credit Scoring

(No exercises)

Module 3: Machine Learning

(No exercises)

Exploratory Data Analysis & Feature Engineering

Module 4: Exploratory Data Analysis

(No exercises here; all exercises are in Module 5)

Module 5: Feature Engineering

• Exercise 1: Exploratory Analysis in R

• Exercise 2: Detection and Treatment of Advanced Outliers

• Exercise 3: Stratified and Random Sampling in R

• Exercise 4: Multivariate imputation model

• Exercise 5: Univariate analysis in percentiles in R

• Exercise 6: Continuous variable optimal univariate analysis in Excel

• Exercise 7: Estimation of KS, Gini, and IV in Excel

• Exercise 8: Word Cloud analysis in R

Machine Learning

Module 6: Unsupervised Models

• Exercise 14: Core components in R and SAS

• Exercise 15: K-Means segmentation in R

Module 7: Logistic Regression & LASSO

• Exercise 16: Logistic Regression in SAS & R

• Exercise 17: LASSO Logistic Regression in R

• Exercise 18: Model risk using confidence intervals

Module 8: Trees, KNN & Naive Bayes

• Exercise 19: Decision Tree in SAS & R

• Exercise 20: KNN in R & SAS

• Exercise 21: Naive Bayes in R

Module 9: Support Vector Machine (SVM)

• Exercise 22: SVM in R (data 1)

• Exercise 23: SVM in Python (data 2)

Module 10: Ensemble Learning

• Exercise 24: Boosting in R

• Exercise 25: Bagging in R

• Exercise 26: Random Forest in R & Python

• Exercise 27: Gradient Boosting Trees

Model Validation

Module 11: Validation of ML & Traditional Models

• Exercise 28: Logistic Regression Goodness-of-Fit Test

• Exercise 29: Cross-validation in SAS

• Exercise 30: Gini, IV, Brier, Lift, CAP, ROC in SAS/Excel

• Exercise 31: Bootstrapping of SAS parameters

• Exercise 32: Jackknifing in SAS

• Exercise 33: ROC/Gini Bootstrapping in SAS

• Exercise 34: Kappa estimation

• Exercise 35: K-Fold Cross-Validation in R

• Exercise 36: Traffic-light validation (6-year horizon)

Module 12: Stability Testing

• Exercise 37: Stability tests of models and factors

Deep Learning

Module 14: Introduction to Deep Learning

(No exercises)

Module 15: Deep Learning Feed Forward Neural Networks

• Exercise 38: Deep Learning Feed Forward Credit Scoring

Module 16: Convolutional Neural Networks (CNN)

• Exercise 39: Credit Scoring using CNN

Module 17: Recurrent Neural Networks (RNN / LSTM)

• Exercise 40: Credit Scoring using LSTM

Module 18: Generative Adversarial Networks (GANs)

• Exercise 41: Credit Scoring using GANs

Module 19: Calibration of ML & DL Models

• Exercise 42: Optimization — XGBoost, Random Forest & SVM

• Exercise 43: Optimized Deep Learning Credit Scoring

Scorecard Construction

Module 20: Traditional Scorecard Construction

• Exercise 44: Building Scorecard (Excel, R, Python)

• Exercise 45: Optimum cut-off point estimation

• Exercise 46: Confusion matrix: Type I & Type II error

Quantum Machine Learning

Module 21: Quantum Credit Scoring

• Exercise 47: Quantum K-means

• Exercise 48: Quantum SVM

• Exercise 49: Quantum Feed Forward NN

• Exercise 50: Quantum Convolutional NN

Module 22: Tensor Networks for QML

• Exercise 51: Credit scoring with tensor networks

Probabilistic Machine Learning

Module 23: Probabilistic Machine Learning

• Exercise 52: Gaussian Process Regression

• Exercise 53: Bayesian Neural Networks Credit Scoring

Model Risk

Module 24: Model Risk in Credit Scoring

• Exercise 54: Model risk due to not recalibrating

Credit Scoring Applications

Module 25: Credit Scoring Models by Product

• Exercise 55: Behavior Score Logistic Regression (Python)

• Exercise 56: Behavior Score SVM (Python)

• Exercise 57: Behavior Score Random Forest (Python)

• Exercise 58: Behavior Score Gradient Boosting (Python)

• Exercise 59: Behavior Score LSTM (Python)

Module 26: Typology of Scoring Models

• Exercise 60: Fraud Score with Neural Networks

• Exercise 61: Income Score

• Exercise 62: Collection Score

• Exercise 63: Recovery Score

• Exercise 64: Churn/Quit Score

Calibration of PD Models

Module 27: PD Calibration (IRB)

• Exercise 65: PD Calibration in ML Models

Module 28: Machine Learning for Lifetime PD (IFRS 9)

• Exercise 66: PD Lifetime — Logistic Regression

• Exercise 67: PD Lifetime — Multinomial Regression

• Exercise 68: PD Lifetime — SVM

• Exercise 69: PD Lifetime — Deep Learning

• Exercise 70: PD Lifetime — LSTM

Validation of PD Models

Module 29: Validation of PD Models

• Exercise 71: Backtesting PD in Excel

• Exercise 72: Forecasting PD vs Real PD

Automation of Credit Scoring & PD

Module 30: Automation of Modeling with AI

• Exercise 73: Automated scoring — modeling, optimization & validation

• Exercise 74: Automation of PD modeling & validation

Module 1: Foundations of Generative AI – Prompts and Embeddings in Credit Risk

• Exercise 1: Prompt a Credit Risk Memo with OpenAI or a local model

• Exercise 2: Generate and visualize text embeddings (PCA or t-SNE on borrower narratives)

Module 2: Foundations of Generative AI and AI Agents in Finance

• Exercise 3: Generate synthetic borrower names and profiles using LLMs

• Exercise 4: Simulate a simple agent making approve/reject decisions based on rules

• Exercise 5: Create a credit applicant description using a pre-trained language model

Module 3: Generative Adversarial Networks (GANs) for Credit Data Synthesis

• Exercise: Build a basic GAN in PyTorch or TensorFlow for credit data

• Exercise: Use t-SNE/PCA to visualize real vs. synthetic borrowers

• Exercise: Train a logistic regression on GAN-augmented data and evaluate AUC

Module 4: Variational Autoencoders (VAEs) and Latent Credit Risk Modeling

• Exercise 6: Implement a VAE to reduce credit scoring data from 20 to 5 dimensions

• Exercise 7: Cluster latent borrower profiles to segment high-risk borrowers

• Exercise 8: Generate 100 new synthetic applicants using decoder manipulation

Module 5: LLMs for Generating Credit Narratives and Reports

• Exercise 9: Convert credit scores, income, and history into a narrative with GPT-3 or DeepSeek

• Exercise 10: Score generated memos using sentiment and readability metrics

• Exercise 11: Create a prompt toolkit for different reporting formats (memo, summary, compliance)

Module 6: Reinforcement Learning for Credit Policy Optimization

• Exercise 12: Implement a Q-learning agent for loan approvals (accept/reject)

• Exercise 14: Introduce penalty for wrong approvals and reward for correct rejections

• Exercise 15: Compare RL agent vs. logistic model on profit-maximizing loan strategy

Module 7: AI Agents for Credit Workflow and Decision Automation

• Exercise 16: Create an AI underwriting assistant using LangChain

• Exercise 17: Give the agent access to a scoring model and a document retrieval tool

• Exercise 18: Simulate 5 credit applications and record the agent’s reasoning paths

Module 8: Explainability and Fairness in Generative AI Models

• Exercise 19: Generate SHAP values for both real and synthetic data models

• Exercise 20: Compare fairness metrics before and after synthetic augmentation

• Exercise 21: Generate an XAI report on credit scoring using LLM-generated explanations

Module 9: Stress Testing Credit Models with Synthetic Scenarios

• Exercise 22: Use an LLM to generate 3 plausible adverse macroeconomic scenarios

• Exercise 23: Simulate portfolio default rates under synthetic shocks

• Exercise 24: Create a stress dashboard using Plotly/Dash

Module 10: Multi-Agent Simulations of Credit Portfolios

• Exercise 25: Build an agent-based model (ABM) for 1,000 agents with different risk propensities

• Exercise 26: Simulate policy interventions (payment holidays, moratoria)

• Exercise 27: Plot heatmaps of default clusters under different scenarios

Module 11: Deploying Generative AI and Agents in Credit Risk

• Exercise 27: Deploy a credit scoring model as a REST API

• Exercise 28: Set up logging of LLM decisions and audit trails

• Exercise 29: Create a compliance checklist for model risk governance

Module 12: Credit Rating using Structured and Unstructured Data with Generative AI

• Exercise 30: Credit Rating – Financial + News Fusion Model (structured financials + text embeddings + classification; compare AUC with and without text)