You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As a followup to #45190 , add a normal rank (or gaussian rank) function that computes the normal percent-point function of the quantile rank results.
In essence, if quantile rank returns quantiles between 0 and 1, gaussian rank returns the points on the normal distribution corresponding to those quantiles.
Component(s)
C++
The text was updated successfully, but these errors were encountered:
### Rationale for this change
Computing ranks as values of the "probit" function (https://en.wikipedia.org/wiki/Probit), rather than quantiles between 0 and 1, can be useful for machine learning and other tasks.
### What changes are included in this PR?
Add a "rank_normal" function that computes array ranks as points on the normal distribution.
It is similar to calling the "rank_quantile" function and then applying the normal percent-point function ("probit").
### Are these changes tested?
Yes, by dedicated unit tests.
### Are there any user-facing changes?
No, except a new compute function.
* GitHub Issue: #45572
Authored-by: Antoine Pitrou <antoine@python.org>
Signed-off-by: Antoine Pitrou <antoine@python.org>
Describe the enhancement requested
As a followup to #45190 , add a normal rank (or gaussian rank) function that computes the normal percent-point function of the quantile rank results.
In essence, if quantile rank returns quantiles between 0 and 1, gaussian rank returns the points on the normal distribution corresponding to those quantiles.
Component(s)
C++
The text was updated successfully, but these errors were encountered: