MLOps & Risk Professional 11 min

Advanced Risk Scoring: Gradient Boosting, Fairness, and Production

Machine learning methods applied to credit scoring have shifted from theoretical research to standard practice. The models in this program reflect that shift — built from data patterns, validated on real portfolios, and interpreted with tools that regulators and risk teams actually use.

5 weeks duration
8! seats left
4.5 / 5 avg rating (105 reviews)
Advanced Risk Scoring: Gradient Boosting, Fairness, and Production program cover
2,190 CAD
Price includes six months of access to updated materials as regulations evolve
enroll now
only 8 spots remaining
format Professional
duration 5 weeks
category MLOps & Risk
reading time 11 min
price 2,190 CAD

about this program

Building a model that scores well on a holdout set is one problem. Keeping it reliable in production, compliant with fair lending rules, and maintainable by a team of three is a different problem entirely.

The production gap

Many data scientists can train a LightGBM model. Far fewer have dealt with score instability after a macroeconomic shift, or had to explain a model rejection to a regulator. Both scenarios are covered here with documented case walkthroughs.

Fairness and disparate impact

Canadian and US fair lending frameworks require that automated decisions do not produce disparate impact on protected groups — even unintentionally. This course covers statistical fairness metrics, disparate impact testing, and how to adjust feature selection and thresholds without gutting predictive power.

Fairness constraints and model performance are not always in opposition. The course demonstrates cases where fairer models also generalize better out of sample.

MLOps for scoring pipelines

You will set up a basic MLflow tracking server, log model versions, and build a PSI-based monitoring script that flags score distribution shifts. The tooling is kept minimal and reproducible — no cloud vendor lock-in.

4.5
based on 105 reviews
what you receive
  • certificate of completion
  • model code and datasets
  • access to Q&A sessions
  • peer review exercises

program curriculum

structured by topic period and learning outcome

foundation
Credit data fundamentals
Working with bureau data, application data, and behavioural variables. Data quality checks and missing-value treatment.
Variable selection and encoding
Weight of evidence encoding, information value, coarse classification — the tooling that still dominates production environments.
model development
Logistic regression as baseline
Why lasso-regularised logistic regression remains a strong default. Scorecard conversion and odds calibration.
Gradient boosting for scoring
XGBoost and LightGBM applied to imbalanced credit datasets. Hyperparameter search without data leakage.
Survival models and through-the-door populations
Handling sample bias from reject inference. Survival analysis for time-to-default outcomes.
validation and deployment
Model performance metrics
Gini coefficient, KS statistic, PSI, and the distinction between discrimination and calibration in regulatory contexts.
Explainability with SHAP
Producing audit-ready explanations for decisions made by tree ensembles. SHAP values in practice, not just theory.
Monitoring and drift detection
Tracking population stability post-deployment. Triggers for model rebuild vs recalibration.
  1. Gradient Boosting Deep Dive

    Week 1

    LightGBM and XGBoost internals, monotonicity constraints for regulatory compliance, hyperparameter search with Optuna.

  2. Fairness Auditing

    Week 2

    Disparate impact ratio, equal opportunity difference, testing feature proxies, fairness-aware feature selection.

  3. Explainability at Scale

    Week 3

    SHAP waterfall plots, global feature importance, building automated reason-code generators.

  4. Monitoring and Drift

    Week 4

    PSI and CSI metrics, setting alert thresholds, scheduling retraining pipelines with Prefect.

  5. MLflow and Model Registry

    Week 5

    Experiment tracking, model versioning, staging and production promotion workflow.

questions? write to [email protected]

ready to work with REAL scoring data?

The program runs on actual credit portfolio datasets under controlled conditions. You leave with code, documentation, and a validated model — not slides. Enrolment closes when seats fill.

duration 5 weeks
8 seats remaining
Your data preferences

We use data to keep things running and to improve your quiz experience: