Introduction | A compact unrestricted model | Preparing the refinement | Smoothing an ordered effect | Inspecting the proposal | Refining a selected part of the smoothing | Explicitly redirecting the curve | Applying a relative local adjustment | Adjusting the slope after an anchor | Interpreting the premium effect | Restricting selected levels | Shrinking a categorical effect | Adding differentiation within model levels | Combining and reviewing refinements | Refitting the model | Inspecting the fitted result | Calibrating the final level | Auditing the portfolio effect | When to revisit the model | Iterating without losing the specification | Summary | Where to go next

Introduction | From point exposures to reporting areas | Aggregate exposure by polygon | Create a choropleth | Interpreting the map | Relation to fixed-radius concentration | Summary

Motivation | Example portfolio | Quick example: find the largest concentration | Understanding the result | Inspecting contributing locations | Calculating sums at specified locations | Continuous vs observed centres | Multiple hotspots | How the search works | Polygon reporting | Practical considerations | Mathematical background | Reference

Introduction | Portfolio and data | Understanding rating factors | Factor analysis | Basic factor analysis | Visualising factor behaviour | Handling continuous variables | Why continuous variables are treated separately | Analysing the shape with a GAM | Deriving tariff segments | Adding tariff segments to the data | Estimating frequency and severity | Why GLMs are used | Frequency model | Severity model | Constructing technical risk premium | From technical risk to tariff | Fitting a tariff representation | Interpreting model effects and observed experience | Rating table | Visualising coefficients | Validating the model | Overdispersion | Model performance | Bootstrap performance | Refining the tariff | Summary | Where to go next

Why large portfolios require a different workflow | Memory rather than a fixed row limit | From policy records to model points | Optional temporal consolidation | Local model-point aggregation | Aggregation and GLM estimation | Lazy model-point aggregation with DuckDB | Scaling the same workflow to 10 and 50 million rows | Consolidating policy periods in DuckDB | Which reduction comes first? | Where to go next

What does model validation mean? | Validation map | Compact example setup | Comparing model performance | Inspecting tariff structure and observed experience | Assessing resampling stability | Checking distributional assumptions | Overdispersion | Simulation-based residuals | Reviewing observed versus expected experience | Validation, investigation and refinement | Putting the evidence together | Where to go next

Package map | Understanding the portfolio | Preparing modelling data | Treating large losses | Understanding and structuring risk factors | Estimating technical risk | Interpreting model effects | Refining the tariff | Validating the model | Scaling to large portfolios | Choosing the building blocks | Where to go next