investr: An R Package for Inverse Estimation

Introduction | Surrogate residuals | Jittering for general models | Bootstrapping | Surrogate residuals in R | Detecting a misspecified mean structure | Obtain the SBS/probability-scale residuals | Residual-vs-covariate plot and Q-Q plot | Detecting heteroscedasticty | Detecting a misspecified link function | Checking the proportionality assumption | Detecting interaction effects | Bitterness of wine | References

Local explanations | Raw Shapley values and Monte Carlo uncertainty | Global explanations | Exact vs. approximate Shapley values | An alternative Monte Carlo estimator: permutation walks | Main, total, and interaction effects | Parallel processing | References

Introduction | Prerequisites | How statlingo Works: The explain() Function and ellmer | Understanding explain()'s Arguments | The Power of context: Why It Matters | Some Examples in Action! | Example 1: Linear Regression (lm) - Sales of Child Car Seats | 1. Model Appropriateness and Research Question | 2. Model Specification | 3. Residuals Summary | 4. Coefficients Table | 5. Residual Standard Error | 6. R-squared Values | 7. F-statistic | 8. Suggestions for Checking Assumptions | Caution | Follow-up Question: Interpreting R-squared | Example 2: Logistic GLM (glm) - Pima Indians Diabetes | Core Concepts & Purpose | Key Assumptions | Assessing Model Appropriateness | Interpretation of the GLM Output | Call / Model Specification | Deviance Residuals (Not provided in summary, but generally) | Coefficients Table | Dispersion Parameter | Deviance Statistics | AIC (Akaike Information Criterion) | Number of Fisher Scoring iterations | Suggestions for Checking Assumptions | Additional Considerations | Example 3: Cox Proportional Hazards Model (coxph) - Lung Cancer Survival | Cox Proportional Hazards Model Interpretation: Lung Cancer Survival | Key Findings | Overall Model Performance | Data Summary | Important Considerations for Next Steps | Example 4: Linear Mixed-Effects Model (lmer from lme4) - Sleep Study | Requesting Plain Text Output (style = "text") | Requesting JSON Output (style = "json") | Suggesting Next Coding Steps | Inspecting LLM Interaction | Conclusion

Example data | Model-specific importance | Permutation importance | Variance-based importance (FIRM) | Shapley-based importance | Plotting options | References

Installation | A first example | Two predictors | 3-D surfaces and three predictors (lattice) | Classification | Models with non-Gaussian responses | Controlling the grid | Learn more

Batched predictions (batch.size) | Parallel processing | The recursive method for gbm | Fast approximate PDPs (approx = TRUE)

ICE curves | Centered ICE (c-ICE) curves | User-supplied prediction functions | References