auditor review
A DALEX-based R package for checking model fit, residuals, performance, and influence.
Reviewed by iTechGuides Editors · Editorial team · Updated Oct 2026
auditor is an open-source R package from MI² DataLab for model-agnostic validation and error analysis of predictive models. It is aimed at people working with models in R who need to examine fit, residuals, performance, or influential observations. The package works with DALEX explainer objects and covers diagnostics for regression and classification. It is programmatic software, installed from CRAN or GitHub, rather than a hosted service. Choose it when the task is model checking within an R and DALEX workflow; it is not presented as a general-purpose prediction-explanation platform.
Its diagnostic range is centered on fit and errors: half-normal plots support model-fit checks, while residual plots and scores help examine prediction errors. It can calculate accuracy, AUC, F1, MAE, and RMSE, compare model performance, and display model rankings. Predicted responses can be visualized against observed or variable values, and Cook’s distances help identify influential observations. These capabilities make the package relevant to reviewing how predictive models perform and where observations may affect them, rather than to explaining individual predictions through feature contributions.
The package is open source with a free plan and supports Linux, Windows, and macOS; deployment is self-hosted. The supplied integration is DALEX, and documentation is the listed support channel. Its scope is a meaningful boundary: feature attribution, counterfactuals, fairness analysis, and audit reports are not listed as capabilities. Teams seeking those functions in an explainability or audit product should consider alternatives. For R practitioners who already work with DALEX and want programmatic diagnostics and model comparisons, auditor is a focused fit; users seeking a hosted interface or broader explanation and governance workflows should look elsewhere.
auditor pros and cons
- Where it wins
- Checks model fit with half-normal plots and residual diagnostics.
- Computes performance metrics including accuracy, AUC, F1, MAE, and RMSE.
- Compares models and uses Cook’s distances to examine influential observations.
- Where it doesn't
- Does not list feature-attribution functionality.
- Does not list counterfactual functionality.
- Does not list fairness analysis or audit reports.
auditor fact sheet, pricing and score →
Advertiser disclosure: iTechGuides is reader-supported. Vendors can pay for top positions in our rankings and for a place on other products' pages, and we may earn a commission when you click some links. How we rank.
Last updated · How we research and update