objects.Text: Is it possible to center texts for stacked bar plot?

[frfeng]

Hi,

For the objects interface is it possible to center the texts inside the stacked bars? See the link below for examples. Thank you.

https://stackoverflow.com/questions/41296313/stacked-bar-chart-with-centered-labels

MWE:

import pandas as pd
import seaborn.objects as so

df = pd.DataFrame(data={'a': [45, 17, 47], 
                        'b': [91, 70, 72], 
                        'c': [68, 43, 13],
                        'group': ['x', 'y', 'z'],
                       }
                 )

(
    so.Plot(df.melt(id_vars='group'), 
            x="group", y="value", color="variable", text="value")
      .add(so.Bar(), so.Stack())
      .add(so.Text(color="k", valign="top"), so.Stack())
)

[thuiop]

Currently no this is not possible, and it does not seem like a trivial feature either since this requires knowing where the middle of the bars is when creating the matplotlib text objects. But there may be a way to modify the so.Bar object so that it retrieves the text and calls ax.bar_label...

[thuiop]

Well, I actually needed this myself so I went ahead and coded it ; this requires a custom object replacing Stack but outputting the middle "y" values.

from dataclasses import dataclass

import matplotlib as mpl
import numpy as np
from pandas import DataFrame
from seaborn._core.groupby import GroupBy
from seaborn._core.moves import Move
from seaborn._core.typing import Default

@dataclass
class MiddleStack(Move):
    def _stack(self, df, orient):

        if df["baseline"].nunique() > 1:
            err = "Stack move cannot be used when baselines are already heterogeneous"
            raise RuntimeError(err)

        other = {"x": "y", "y": "x"}[orient]
        stacked_lengths = (df[other] - df["baseline"]).dropna().cumsum()
        offsets = stacked_lengths.shift(1).fillna(0)

        df[other] = stacked_lengths - (df[other]- df["baseline"])/2
        df["baseline"] = df["baseline"] + offsets

        return df

    def __call__(
        self, data: DataFrame, groupby: GroupBy, orient: str, scales: dict[str, Scale],
    ) -> DataFrame:

        groupers = ["col", "row", orient]
        return GroupBy(groupers).apply(data, self._stack, orient)

You can use it by replacing so.Stack by MiddleStack ; don't forget to also remove the valign.

[frfeng]

Thank you for sharing your solution. I got an error when running your code.

---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
Cell In[4], line 11
      7 from seaborn._core.moves import Move
      8 from seaborn._core.typing import Default
     10 @dataclass
---> 11 class MiddleStack(Move):
     12     def _stack(self, df, orient):
     14         if df["baseline"].nunique() > 1:

Cell In[4], line 28, in MiddleStack()
     23     df["baseline"] = df["baseline"] + offsets
     25     return df
     27 def __call__(
---> 28     self, data: DataFrame, groupby: GroupBy, orient: str, scales: dict[str, Scale],
     29 ) -> DataFrame:
     31     groupers = ["col", "row", orient]
     32     return GroupBy(groupers).apply(data, self._stack, orient)

NameError: name 'Scale' is not defined

And for testing when I remove the Scale part in the line 28 above, the error disappears, but I got the following error when calling so.MiddleStack().

     11 (
     12     so.Plot(df.melt(id_vars='group'), 
     13             x="group", y="value", color="variable", text="value")
---> 14       .add(so.Bar(), so.MiddleStack())
     15       .add(so.Text(color="k"), so.MiddleStack())
     16 )

AttributeError: module 'seaborn.objects' has no attribute 'MiddleStack'

Am I doing something wrong? Thank you.

[thuiop]

You should remove the "so.", since the object is not embedded in the seaborn.objects module. Also sorry about the Scale thing, an import is missing ; this is only a type hint anyway so it does not matter much.

[frfeng]

That makes sense. It works now. Thank you! For future reference below is the complete example.

from dataclasses import dataclass

import matplotlib as mpl
import numpy as np
from pandas import DataFrame
from seaborn._core.groupby import GroupBy
from seaborn._core.moves import Move
from seaborn._core.typing import Default

@dataclass
class MiddleStack(Move):
    def _stack(self, df, orient):

        if df["baseline"].nunique() > 1:
            err = "Stack move cannot be used when baselines are already heterogeneous"
            raise RuntimeError(err)

        other = {"x": "y", "y": "x"}[orient]
        stacked_lengths = (df[other] - df["baseline"]).dropna().cumsum()
        offsets = stacked_lengths.shift(1).fillna(0)

        df[other] = stacked_lengths - (df[other]- df["baseline"])/2
        df["baseline"] = df["baseline"] + offsets

        return df

    def __call__(
        self, data: DataFrame, groupby: GroupBy, orient: str, scales: dict[str],
    ) -> DataFrame:

        groupers = ["col", "row", orient]
        return GroupBy(groupers).apply(data, self._stack, orient)


# plotting
import pandas as pd
import seaborn.objects as so

df = pd.DataFrame(data={'a': [45, 17, 47], 
                        'b': [91, 70, 72], 
                        'c': [68, 43, 13],
                        'group': ['x', 'y', 'z'],
                       }
                 )

(
    so.Plot(df.melt(id_vars='group'), 
            x="group", y="value", color="variable", text="value")
      .add(so.Bar(), so.Stack())
      .add(so.Text(color="k"), MiddleStack())
)