What the learner will be able to do
- ✓Build and validate credit-risk models.
- ✓Compare scorecards and machine learning.
- ✓Explain predictions responsibly.
- ✓Design monitoring and governance.
An eight-week live online course in modern credit-risk modelling, validation, explainability and governance.
An eight-week live online course in modern credit-risk modelling, validation, explainability and governance.
Working knowledge of statistics and Python is recommended.
Open a week to inspect its lessons and applied focus. The current foundation pathway contains only the six active curriculum weeks; historical versions remain preserved privately for existing learner progress.
Week 1
Define default, observation windows, outcomes and business use.
Applied work for this week
Students translate a lending objective into a measurable probability-of-default target, define the observation and performance windows, and identify leakage risks. They produce a model-development brief that connects the target, portfolio scope, decision use and business cost of errors.
Week 2
Build representative samples and prevent target leakage.
Applied work for this week
Students profile an applicant and repayment dataset, test missingness and outliers, compare sampling strategies and document every exclusion. They deliver a data-quality report and an analysis-ready sample with a defensible train, validation and out-of-time split.
Week 3
Create interpretable baselines and risk-ranking measures.
Applied work for this week
Students bin candidate variables, assess predictive power and stability, and build an interpretable logistic-regression scorecard. They explain coefficient direction, convert probabilities into scores and test whether the model supports a real credit decision.
Week 4
Compare flexible models against the baseline.
Applied work for this week
Students train decision-tree and gradient-boosting challengers, tune complexity and compare nonlinear lift with the scorecard baseline. They document where additional performance is commercially useful and where it creates unacceptable opacity or instability.
Week 5
Assess discrimination, calibration and stability.
Applied work for this week
Students calculate discrimination, calibration and ranking metrics across development and out-of-time samples. They recalibrate predicted default rates, select thresholds using business costs and write a concise validation conclusion with limitations.
Week 6
Use explanations without overstating causality or certainty.
Applied work for this week
Students use global and local explanation methods to investigate model drivers, adverse decisions and protected-group risks. They prepare an explainability pack that a validator, credit officer and affected customer could each understand.
Week 7
Define controls, thresholds, review cycles and documentation.
Applied work for this week
Students design population-stability, performance and data-quality monitoring with trigger levels and escalation owners. They map model documentation, independent validation, change control and regulatory responsibilities into a practical governance plan.
Week 8
Defend the complete model, limitations and implementation plan.
Applied work for this week
Students assemble an end-to-end credit-risk model, validation evidence, monitoring design and executive recommendation for a realistic lending case. They defend performance, fairness, explainability and business impact, then revise the submission using Synapse Review Coach feedback.
This summary shows the complete learning and assessment trail used for course completion. Certificate eligibility is created only after the required weekly work, knowledge checks and applied assessments are completed. The verified certificate is included with this programme.