PResiduals: An R Package for Residual Analysis Using Probability-Scale Residuals

Area of Science:

• Statistical modeling
• Data analysis software

Background:

• Traditional residual analysis methods have limitations with various outcome types and models.
• Existing diagnostics may not be applicable in complex statistical scenarios.

Purpose of the Study:

• Introduce the PResiduals R package for advanced residual analysis.
• Provide a versatile tool for model diagnostics and association testing.

Main Methods:

• Utilize probability-scale residuals, applicable to a wide range of outcome types and models.
• Implement tests for conditional associations and covariate adjustment for Spearman's rank correlation.
• Develop methods that are robust and efficient for orderable variables without requiring score assignment or transformation.

Main Results:

• The PResiduals package offers a flexible approach to residual analysis.
• Probability-scale residuals are well-defined even when other residuals fail.
• The package facilitates robust and efficient conditional association tests.

Conclusions:

• The PResiduals R package provides a valuable tool for statistical modeling and diagnostics.
• Its probability-scale residuals offer advantages in robustness and applicability.
• The package facilitates advanced data analysis for researchers.