NEWS.md
New vignette “Tidymodels with {equatiomatic}” illustrating that extract_eq() also works with fitted Tidymodels, including with {workflows} objects.
New vignette “Equations in R Markdown and Quarto” that explain how to use the three functions equation(), eq_() and eq__() to ease inclusion of equations in R Markdown and Quarto documents.
extract_eq() can now extract equations from a list of models. There is an example of an application at the end of the “Introduction to {equatiomatic}” vignette.
Proofreading of the README.Rmd/.md files and the vignettes to homogenize style and conventions (e.g., {package} not {package} or package).
Replacement of {magrittr} pipe %>% by the native R pipe |> everywhere and elimination of the dependency to {magrittr} (less dependencies is better).
format.equation() now works for several equations in the equation object (solve issue #207).equation(), eq_() and eq__() ease the inclusion of LaTeX equations in R Markdown or Quarto documents. They also allow to provide the LaTeX code of the equation directly, and they use ‘label’ and ‘units’ attributes from the original dataset automatically to rename the variables in the equations. equation() just returns the LaTeX code. eq_() is meant to easily include inline equation using an inline chunk like r eq_(mod1). eq__() (with two underscores) is meant to be used in a code chunk inside a $$...$$ Markdown construct, possibly with a label ($$...$${#eq-label}) where ... is something like eq__(mod1).preview_eq() uses rmarkdown and pandoc to generate a preview of the equation. The package no longer needs the texPreview package now.Correction in the plotting-integration vignette: ggplot2 now requires a vector for text annotation, so, use as.character() and specify parse = TRUE () PR #244.
The penguins dataset is now explicitly loaded from {equatiomatic} : another version appears in the {datasets} package version 4.5.0 with different column names and it causes some tests to fail and confusion in the examples.
Correction of an URL (https instead of http) in the lme4-lmer vignette.
extract_eq() added for model_fit objects of the {parsnip} package, and for workflow objects of the {workflows} package, see #237. Previously, the model object had to be extracted using parsnip::fit() before using extract_eq(). This is not necessary any more, but still valid (no breaking changes). Note that extract_eq() cannot handle yet all {tidymodels}. For instance, it can handle reg_linear(), but cannot handle reg_linear() |> set_engine("stan"). There is currently no plan to support all {tidymodels} models, but pull requests will be considered if you find and solve a problematic case by yourself.extract_eq() are now properly documented in the extract_eq() help page.add_greek.forecast_ARIMA and add_coefs.forecast_ARIMA (first argument side= is renamed rhs=).for census_division l = 1 is now rendered as for census\_division l = 1).se_subscripts argument, which allows the standard error for each coefficient to be included in parentheses below the coefficient when se_subscripts = TRUE. This is supported for lm and glm models.renderEq() and eqOutput() functions for working with equatiomatic with shiny.greek_colors, subscript_colors, var_colors and var_subscript_colors arguments.swap_var_names and swap_subscript_names arguments to change the names of the variables or subscripts in the rendered equationlme4::glmer() models.return_variances argument, which allows users to optionally return the variance/covariances in lme4::lmer() and lme4::glmer() modelsfont_size argument, which takes any LaTeX font size (see here)label argument, which allows for cross-referencing equation in-text with PDF outputs.poly(), log(), and exp() functions for in-line equations, and will drop the I() for generic in-line equation operationlme4::lmer() and and lme4::glmer() modelslme4::lmer() modelslme4::lmer() models now allow for in-line alterations to the code, e.g., I(n >5).New models: This is the first version to support lme4::lmer() models and forecast::Arima() models.
New vignettes for each of the new models supported.
New vignette showing how to use the package with plotting
Switched the testing framework to use snapshot testing
Uses new print method so users no longer have to specify results = "asis" in the R Markdown chunk option
Fixed error in rendering logistic/probit regression equations by removing the epsilon (error term) at the end of the left-hand side
The epsilon (error term) is no longer shown when rendering the fitted model equation, i.e. use_coefs = TRUE, for lm models. The hat sign is also added to the response variable.
Initial CRAN Release
Extracts equations for lm models
Extracts equations for glm models with family = binomial(link = "logit") or family = binomial(link = "probit")
Extracts equations form ordered regression models using MASS::polr or ordered::clm for logit and probit link functions
glm models have an optional show_distribution argument to show the distributional assumptions
All equations can be displayed using Greek notation or the estimated coefficients
The raw_tex code can be used to supply custom TeX code for the intercept or the coefficients, through the intercept and greek arguments respectively
Long equations can be wrapped to multiple lines with the optional wrap argument. The length of the wrapping is controlled by terms_per_line.
Added a NEWS.md file to track subsequent changes to the package.