First of all, thank you for the great work with these packages, and for implementing changes so quickly.
I may well be too early, testing functions that aren't official yet.
But anyway, this is how insight is working right now.
library(easystats)
packageVersion("insight") # ‘0.99.0.15’
packageVersion("performance") # ‘0.12.4.5’
packageVersion("easystats") # ‘0.7.3.2’
a <- aov(formula = (mpg ^ 1.3 - 1) / 1.3 ~ hp, data = mtcars)
insight::find_transformation(a) # "box-cox" - correct
insight::get_transformation(a) # also has a correct output
a <- aov(formula = (mpg ^ -1.3 - 1) / -1.3 ~ hp, data = mtcars)
insight::find_transformation(a) # "power" - not correct
debugonce(insight::get_transformation)
insight::get_transformation(a)
# Error in !is.null(trans_power) && trans_power != 0 :
# 'length = 2' in coercion to 'logical(1)'
# In addition: Warning message:
# In doTryCatch(return(expr), name, parentenv, handler) :
# NAs introduced by coercion
The error in get:transformation() appears to be caused by the misdirection from find_transformation().
When running .extract_power_transformation(model) instead of .extract_scale_denominator(model) on the formula, we get c(NA, 1.3) instead of the single value that the next line wants to evaluate.
But there may be more to this bug. While in debugging mode inside .get_transformation(), I evaluated .extract_scale_denominator(model), and it becomes 1.3 but I think it ought to be -1.3.
Something happens in the lapply call between line 21 and 31 of insight:::.get_variables_list(), when f$response goes from "(mpg^-1.3 - 1)/-1.3" to "(mpg^-1.3 - 1)" "1.3".
First of all, thank you for the great work with these packages, and for implementing changes so quickly. I may well be too early, testing functions that aren't official yet. But anyway, this is how insight is working right now.
The error in
get:transformation()appears to be caused by the misdirection fromfind_transformation(). When running.extract_power_transformation(model)instead of.extract_scale_denominator(model)on the formula, we getc(NA, 1.3)instead of the single value that the next line wants to evaluate.But there may be more to this bug. While in debugging mode inside
.get_transformation(), I evaluated.extract_scale_denominator(model), and it becomes 1.3 but I think it ought to be -1.3. Something happens in the lapply call between line 21 and 31 ofinsight:::.get_variables_list(), whenf$responsegoes from"(mpg^-1.3 - 1)/-1.3"to"(mpg^-1.3 - 1)" "1.3".