Overview | Installation | The score scale and ncuts | Likelihood and censoring types | Interval construction | Data preparation with brs_prep() | Mode 1: Score only (automatic) | Mode 2: Score + explicit censoring indicator | Mode 3: Interval endpoints with NA patterns | Mode 4: Analyst-supplied intervals | Using brs_prep with brs() | Example 1: Fixed dispersion model | Simulating data | Fitting the model | Estimation and fit diagnostics | Goodness of fit | Comparing link functions | Residual diagnostics | Predictions | Confidence intervals | Censoring structure | Example 2: Variable dispersion model | Accessing coefficients by submodel | Comparing link functions (variable dispersion) | Diagnostics for variable dispersion | Advanced analyst functions | Parametric bootstrap confidence intervals | Average marginal effects (AME) | Score probabilities on the original scale | Repeated k-fold cross-validation | S3 methods reference | Reparameterizations | References

Scope, and how this was written | Two ways to run an algorithm | obwoe() does not forward algorithm-specific parameters | Coverage | The quantity being optimized | Choosing an algorithm | Reference, organized by mechanism | Exact dynamic programming | Recursive entropy partitioning | Bottom-up MDL cost | Statistical merging | Isotonic regression | Density-based | Distribution-based initialization | Divergence-driven | Streaming | Greedy merging under an IV or WoE criterion | Metaheuristic | Measured behavior | Known limitations | References | What could not be verified

Why bin | The quantities | Weight of Evidence | Information Value | Monotonicity | The data | One variable, end to end | Reading the gains table | All variables at once | Screening with obwoe_select() | The bin-level view | Rare levels and bin_cutoff | Choosing an algorithm | Applying the transformation | Exporting to SQL | Preprocessing | Where to go next | References

1. Three scores on the demo data | 2. Credit: bands, tiers and lights | 3. Fraud: tail bands, action tiers and a daily capacity | 4. Churn: anchored tiers, claims and a budget | 5. Credit and churn on the same customers | 6. Why the default rate moved | 7. One score, many segments | 8. Time, treatment, rules and episodes | 9. Production

Why manual coarse classing still matters | A fast selection run | Opening the lab | Proposing bins | Numeric: breaks, merge, split | Categorical: groups, other_to, missing_to | Reading the warnings and the verdict | Accepting, discarding and the blocked path | Choosing the variables | The session summary and the ledger | The spec round trip: a reviewer edits in a spreadsheet | Committing the lab: what changed | Refitting the scorecard | Production: R and SQL follow the manual bins | Governance

1. Why "700 points" means nothing on its own | The scale is one statement | The orientation trap | 2. The two-step regression alignment | 3. Two targets, two conventions, one scale | 4. From logit to points | 5. Do the odds double every PDO? | 6. A tree challenger on the same scale | The swap set | 7. The portfolio view | 8. Rescaling an existing scorecard | 9. A checklist for the model committee | Reference

1. Configuration | 2. Selection and the funnel | 3. The scorecard | 4. Cut-off and strategy | The sweep | The strategy table | 5. Reject inference | 6. Scoring and reasons | 7. Production SQL | 8. Monitoring | The monitoring plan | 9. Deliverables | 10. Before go-live | Appendix: from a database | Next steps | References

Audience and scope | 1) Reproducible simulation and data checks | 2) Fixed-effects candidate set and model ranking | 3) Inference stack: Wald + bootstrap + AME | 4) Prediction layer on analyst scale | 5) Out-of-sample validation | 6) Escalation to mixed models | 6.1 Simulate clustered data with random intercept + slope | 6.2 Fit evolutionary sequence | 6.3 Model choice by LLR/LRT (ANOVA) | 7) Random-effects study (numeric + visual) | 8) A Monte Carlo study (design of Lopes, 2023, ch. 3) | 9) Practical decision checklist | References

Overview | Mathematical foundations | Complete likelihood by censoring type | Model-comparison metrics | Average marginal effects (AME) | Score-scale probabilities | Cross-validation log score | Reproducible workflow | 1) Simulate data and fit candidate models | 2) Compare models in one table | 3) Estimate average marginal effects | 4) Predict score probabilities | 5) ggplot2 diagnostics | 6) Repeated k-fold cross-validation | Practical interpretation | References

Overview | Mathematical model | Linear predictors | Beta parameterization | Conditional contribution by censoring type | Random-effects distribution | Group marginal likelihood | Laplace approximation used by brsmm() | Other integration methods and estimation details | Simulating clustered scale data | Fitting brsmm() | Random intercept + slope example | Additional studies of random effects (numerical and visual) | Core methods | Coefficients and random effects | Variance-covariance, summary and likelihood criteria | Fitted values, prediction and residuals | Diagnostic plotting methods | Prediction with newdata | Statistical tests and validation workflow | Wald tests (from summary) | Evolutionary scheme and Likelihood Ratio (LR) test selection | Residual diagnostics (quick checks) | Parameter recovery experiment | How this maps to automated package tests | References

Introduction | Key Features | The Distribution Family | Mathematical Foundation | Nested Sub-families | Basic Distribution Functions | Density, CDF, Quantile, and Random Generation | Visualization | Comparing Distribution Shapes | Flexibility Across Families | Maximum Likelihood Estimation | Basic MLE Workflow | Inference Results | Goodness-of-Fit Assessment | Model Selection | Comparing Nested Models | Interpretation | Advanced Topics | Profile Likelihood | Confidence Regions | Performance Benchmarking | Computational Efficiency | Practical Applications | Example: Modeling Proportions | Recommendations | When to Use Each Distribution | Model Selection Workflow | Conclusion | Further Reading | Session Information

Abstract | Introduction | Motivation and Background | The Kumaraswamy Alternative | Study Objectives | Theoretical Framework | Distributional Families | Beta Distribution | Kumaraswamy Distribution | Exponentiated Kumaraswamy Distribution | Regression Structures | Parameter Correspondence and Interpretation | Simulation Study | Design and Methodology | Performance Metrics | Scenario 1: Beta-Distributed Data | Visualizing the Response Distribution | Simulation Results | Parameter Estimation Comparison | Scenario 2: Heavy-Tailed Data | Visualizing Heavy-Tailed Distributions | Parameter Estimation Under Misspecification | Scenario 3: Extreme Distributional Shapes | Visualizing Extreme Boundary Concentrations | The Convergence Crisis | Results and Discussion | Comparative Performance Across Scenarios | Statistical Performance Summary | Parameter Estimation: Magnitude and Precision | Computational Efficiency Analysis | Practical Decision Framework | Conclusions | Summary of Findings | Theoretical Implications | Methodological Contributions | References | Session Information

1. Introduction and Preliminaries | 1.1 Motivation and Background | 1.2 Mathematical Preliminaries | 2. The Generalized Kumaraswamy Distribution and Its Subfamilies | 2.1 Definition and Fundamental Properties | 2.2 The Hierarchical Structure | 2.2.1 Beta–Kumaraswamy distribution | 2.2.2 Kumaraswamy–Kumaraswamy distribution | 2.2.3 Exponentiated Kumaraswamy distribution | 2.2.4 McDonald distribution | 2.2.5 Kumaraswamy distribution | 2.2.6 Beta distribution | 3. Likelihood-Based Inference | 3.1 The Log-Likelihood Function | 3.2 Maximum Likelihood Estimation | 3.3 Likelihood Ratio Tests | 4. Analytical Derivatives and Information Matrix | 4.1 The Score Vector | 4.2 The Hessian and Observed Information Matrix | 4.2.1 Diagonal elements | 4.2.2 Off-diagonal elements | 4.3 Asymptotic Variance–Covariance Matrix | 5. Computational Aspects and Discussion | 5.1 Numerical Stability | 5.2 Optimization | 5.3 Gradient Accuracy | 5.4 Practical Recommendations | 5.5 Discussion | References

The setting | A realistic origination base | Missing values become levels | Screening at scale | The two rejections worth reading | Evidence for the committee | Redundancy | The pipeline as a recipe | The model | Scorecard points | Validation | Rank ordering out of time | Population stability | Deployment | To the warehouse | To R | Tuning with tidymodels | Governance checklist | See also