ENH: Series between inclusive end point bifurcation

[colinalexander]

Allow ability to distinguish inclusiveness of left and right boundaries. I've written the code and tests, but need a GH tracking number prior to submitting a PR.

    def between(self, left, right, inclusive=True):
        """
        Return boolean Series equivalent to left <= series <= right. NA values
        will be treated as False

        Parameters
        ----------
        left : scalar
            Left boundary
        right : scalar
            Right boundary
        inclusive : boolean for both left and right or iterable pair.
            Whether or not to include the left and right boundary.
            If a tuple pair, the boundaries can be set separately
            (e.g. (False, True) for left < series <= right).

[jreback]

this should not be a separate method, maybe as an argument to .loc. Though we already follow the conventions for positional indexing, and label indexing by definition includes the endpoints (as its a bit not-trivial not too).

So what exactly is the use case here?

[colinalexander]

Desired functionality is to augment existing functionality of pd.Series.between as follows:

Instead of this...

s = pd.Series(range(5))

>>> s.loc[(s > 2) & (s <= 4)]
3    3
4    4
dtype: int64

Have something like this:

>>> s.between(2, 4, inclusive=(False, True))
3    3
4    4
dtype: int64

Here is the first proposed revision:

    def between(self, left, right, inclusive=True):
        """
        Return boolean Series equivalent to left <= series <= right. NA values
        will be treated as False

        Parameters
        ----------
        left : scalar
            Left boundary
        right : scalar
            Right boundary
        inclusive : Boolean to indicate whether or not to include both the left 
            and right endpoints (True: a <= series <= b, False: a < series < b),  
            or an iterable boolean pair to set them separately, e.g            
            inclusive= (False, True) for a < series <= b.

        Returns
        -------
        is_between : Series
        """
        if inclusive:
            try:
                pair = iter(inclusive)
                left_inclusive = pair.next()
                rigt_inclusive = pair.next()
                lmask = self >= left if left_inclusive else self > left
                rmask = self <= right if rigt_inclusive else self < right
            except TypeError:
                lmask = self >= left
                rmask = self <= right
        else:
            lmask = self > left
            rmask = self < right

        return lmask & rmask

[jorisvandenbossche]

@jreback yes, this method already exists ... (didn't know that! so my first reaction was the same as yours)

So given that, I don't see any harm is extending its functionality a little bit

[jreback]

I suppose ok, but let's deprecate this method entirely. This is just another (confusing) way of doing things. There are already too many ways of doing indexing. This (although a nice name) is just bloating the API.

Pls create another issue for this deprecation (which will have to wait for 0.19.0)

[shoyer]

+1 for deprecating between

[jreback]

closed in favor of #13027