fx_default_fxGeom_trans: Effex Functions: Default of 'fxGeom_trans'
In sflippl/tectr: User Interface for Complex Datasets

Description Usage Arguments Mechanism See Also Examples

The default transformation is determined in a manner that is sensitive to skewness. Skewed distribution are often hard to interpret if the axes are not transformed.

1
2
3

fx_default_fxGeom_trans(data, mf = metaframe(fx_default(data)),
  fxGeom_trans_simple = TRUE, fxGeom_trans_p.threshold = 0.01,
  custom_fun = default_identity)

`data`	A data frame
`mf`	The metaframe. The metaframe of `data` is set as default.
`fxGeom_trans_simple`	`logical` should all boxcox and modulus transformations be allowed or should they be restricted to "identity", "sqrt" and "log10"?
`fxGeom_trans_p.threshold`	a value between 0 and 1. How low does the p-value of the `moments::agostino.test()` have to be to make another transformation necessary? A rather low value is recommended.

This function has two different modi: fxGeom_trans_simple = TRUE/FALSE. TRUE means that the only possible transformations are "identity", "sqrt" and "log10". FALSE, on the other hand, allows for the entire range of boxcox transformations.

Non-numeric data is always assigned "identity".

If fxGeom_trans_simple is true, the skewness of the three transformation is compared and the lowest absolute skewness wins. If there is nonpositive numeric data, "identity" is chosen by default.

If fxGeom_trans_simple is false, the procedure consists of severals steps:

To avoid unnecessary transformations, the moments::agostino.test() for unnormal skewness is conducted. If the corresponding p-value is not below fxGeom_trans_p.threshold, the transformation is "identity".
Afterwards, geoR::boxcoxfit() determines a suitable transformation which is then transformed into a suitable object by scales::boxcox_trans()

scales::boxcox_trans()

set.seed(1)
norm <- rnorm(1000)
square <- norm^2
exp <- exp(norm)
df <- dplyr::tibble(
  norm = rnorm(100),
  square = rnorm(100)^2,
  exp = exp(rnorm(100)),
  chars = rep("a", 100)
)
fx_default_fxGeom_trans(df)
fx_default_fxGeom_trans(df, fxGeom_trans_simple = FALSE)