Course Objective & Scope
Risk scores must be both accurate and explainable: a declined customer or an auditor must be given a reason. The programme covers target and performance-window definition, binning, variable analysis with WoE and Information Value, building a logistic regression scorecard and calibrating scores to PD.
Explainable use of gradient boosting with monotonic constraints and SHAP, reject inference and cut-off strategy are also addressed. Using tools such as optbinning, participants produce a scorecard and score distribution analysis on data similar to their own portfolio.
Who Should Attend?
- Credit risk and scoring teams
- Insurance actuarial and underwriting analytics specialists
- Supplier and counterparty risk analysts
- Prerequisite: basic statistics and Python or SAS/R experience
Key Takeaways
- Define the target and observation/performance windows correctly
- Analyse variables with WoE/IV and apply monotonic binning
- Build a logistic regression scorecard and scale it to points
- Make gradient boosting explainable with monotonic constraints
- Propose a cut-off strategy balancing approval rate and losses
Curriculum
Risk Problem and Target Definition
- Default definition, observation and performance windows
- Vintage analysis and sample design
Binning, WoE and Information Value
- Monotonic binning with optbinning
- Variable selection and correlation checks
Scorecard Development
- Logistic regression, coefficient interpretation and points scaling
- Calibrating scores to PD
ML-Based Risk Models
- Monotonic-constrained LightGBM and SHAP explanations
- Comparing performance and explainability with the scorecard
Reject Inference and Cut-off Strategy
- Inference methods for declined applications
- Thresholds balancing approval rate, losses and revenue
Other Courses in This Category
-
Churn, purchase and cross-sell propensity models covering target definition, customer behaviour features, survival analysis, uplift modelling and campaign targeting, turning predictions into measurable retention and sales actions.
- Live Course
- 5 modules
-
Collaborative filtering, content-based and hybrid recommenders, matrix factorisation, implicit feedback, cold-start strategies and ranking metrics for building personalised product and content recommendation engines.
- Live Course
- 5 modules
-
Failure probability and remaining useful life (RUL) prediction from sensor, telemetry and maintenance logs, with hands-on signal feature engineering, health indices, early-warning thresholds and maintenance cost optimisation.
- Live Course
- 5 modules
-
SKU-location demand forecasting covering promotions, price elasticity, stockout censoring adjustment, intermittent demand (Croston/ADIDA), hierarchical reconciliation and probabilistic forecasts tied to safety stock.
- Live Course
- 5 modules
-
Bridging predictive models and mathematical optimisation: LP/MILP formulation, capacity, scheduling, routing and budget/price allocation with OR-Tools and Pyomo, plus stochastic and contextual decision systems under uncertainty.
- Live Course
- 5 modules
-
Mastering tabular gradient boosting with XGBoost, LightGBM and CatBoost: tree-growth strategies, native categorical and missing-value handling, ranking/quantile/custom objectives, monotonic constraints, Optuna and fast inference.
- Live Course
- 5 modules
Get in touch
