Let's say I have a continuous response and I'm interested in correlations all along its range, not just the top or bottom quartile.
Instead of running a separate correlation_funnel for each bin, is there any reason I cannot add the continuous column back in between running binarize() and correlate()?
The above runs with one warning and no errors, producing a plot where presumably all correlations are relative to an outcome variable that is not binned just like I want.
My question: Is it valid to use correlationfunel this way?
Let's say I have a continuous response and I'm interested in correlations all along its range, not just the top or bottom quartile.
Instead of running a separate correlation_funnel for each bin, is there any reason I cannot add the continuous column back in between running
binarize()
andcorrelate()
?Reprex:
The above runs with one warning and no errors, producing a plot where presumably all correlations are relative to an outcome variable that is not binned just like I want.
My question: Is it valid to use correlationfunel this way?
Thanks.