LogisticComparison

Logistic-model comparison family

Description

LogisticComparison implements nested, heterogeneity, group, and pairwise comparison workflows for binary outcomes using stats::glm().

Summary rows follow associate.schema.comparison.

Super classes

PolyGenius::ComparisonFamily -> PolyGenius::LinearComparison -> LogisticComparison

Methods

Public methods

Inherited methods

Method new()

Register the canonical family key for logistic comparisons.

Usage
LogisticComparison$new()
Returns

The initialized family object with family.name = “glm”.


Method prepare.frame()

Build and normalize the binary comparison frame for one logistic comparison cell.

Usage
LogisticComparison$prepare.frame(frame, cell, options)
Arguments
frame

Resolved analysis frame restricted to the current fit cell.

cell

Named list describing the current fit-grid cell.

options

Normalized comparison options from the engine.

Returns

A plain data frame with the outcome normalized to 0/1.


Method fit.comparison()

Fit the base and full logistic models required by the requested comparison.

Usage
LogisticComparison$fit.comparison(frame, cell, options, ...)
Arguments
frame

Prepared comparison frame returned by prepare.frame().

cell

Named list describing the current fit-grid cell.

options

Normalized comparison options from the engine.

Additional fitting arguments passed to stats::glm().

Returns

A named list of fitted logistic model objects.


Method build.summary.rows()

Convert fitted logistic comparison objects to shared comparison-schema rows.

Usage
LogisticComparison$build.summary.rows(fit, frame, cell, options)
Arguments
fit

Fitted comparison object returned by fit.comparison().

frame

Prepared comparison frame.

cell

Named list describing the current fit-grid cell.

options

Normalized comparison options from the engine.

Returns

A data.table of rows following associate.schema.comparison.


Method clone()

The objects of this class are cloneable with this method.

Usage
LogisticComparison$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.