Let’s try and keep this on topic - most replies to this message has been about #9211, which is an orthogonal issue.

There are two main questions here:

  1. Would the community prefer to use np.quantile(x, 0.25) instead of np.percentile(x, 25), if they had the choice
  2. Is this desirable enough to justify increasing the API surface?

The general consensus on the github issue answers yes to 1, but is neutral on 2. It would be good to get more opinions.


On Fri, 21 Jul 2017 at 16:12 Chun-Wei Yuan chunwei.yuan@gmail.com wrote:

There's an ongoing effort to introduce quantile() into numpy.  You'd use it just like percentile(), but would input your q value in probability space (0.5 for 50%):

Since there's a great deal of overlap between these two functions, we'd like to solicit opinions on how to move forward on this.

The current thinking is to tolerate the redundancy and keep both, using one as the engine for the other.  I'm partial to having quantile because 1.) I prefer probability space, and 2.) I have a PR waiting on quantile().


NumPy-Discussion mailing list