Closed grahamgill closed 5 years ago
Shouldn't values be left unrounded and the tolerance setting of compare_df() used to pick what counts as a difference, no matter the scale
That's a very smart idea - Hadn't thought of it. Will make an update soon regarding this.
I didn't implement rounding as inferred from tolerance because tolerance could be of two types and it gets complicated. I have added a simpler (albeit less elegant) approach of adding a parameter.
@grahamgill - I've pushed a new commit. Can you verify if the new parameter solves the problem? You'll have to install this dev version directly from github using devtools::install_github()
compareDF is very useful, thanks.
What's the purpose of the r2two() function, and rounding doubles to 2 decimal places? Shouldn't values be left unrounded and the tolerance setting of compare_df() used to pick what counts as a difference, no matter the scale?
I want to use the data in compare_df(...)$comparison_df for further computation, not just display or summarising changes. I need the reported values to be what they were originally.
I'm working around this at the moment by multiplying doubles columns by 10^n, with whatever n is relevant to my use case, before calling compare_df(), then converting comparison_df back.