detect_transmutation: Detect transmutation in formulas, such as factor(), and raise...

In regressinator: Simulate and Diagnose (Generalized) Linear Models

Detect transmutation in formulas, such as factor(), and raise an error

Description

We rely on predictors to occur in models with only one type (such as numeric or factor), but the use of factor() would make it possible for a predictor to appear both as a factor or as numeric. The use of factor() also makes it harder to correctly detect the types of predictors, since the methods for obtaining model predictors provide them before they are converted to factor, not after. So we reject formulas that transmute types inside the formula, such as with factor().

Usage

detect_transmutation(formula, call = parent.frame())

Arguments

formula	Model formula
call	Environment in which to raise the error, defaulting to the calling environment. As this function is recursive, this reduces the complexity of backtraces.

Details

Presently only factor() calls are rejected, but if other transmutations (such as conversions to logical or numeric) prove to be problems, they can be rejected as well.

Value

No value. Raises an error if transmutation is present.

regressinator documentation built on Sept. 11, 2024, 6:50 p.m.