Plotting negative and positive contributions in stacked bar plot independently

[FabianHofmann]

In the object interface API, I am struggling to find a good approach on plotting stacked bar plots from positive and negative values at the same time. As far as I see, all contributions add on top of each other. This means that if a negative value follows a positive, it will start from the end point of the positive, resulting in automatic overlapping.

Is there a possible way to disable overlapping of contributions in stacked bar plots? Like this negative values would always stack on the negative side, positives values on the positive side.

Many thanks for the great package!

[mwaskom]

It's a little unclear what you're trying to do here. Can you show an example using one of the built-in datasets?

[FabianHofmann]

@mwaskom thank you for being so responsive. Let me try to be a bit more precise. Here is an example of a dataframe with positive and negative values. When plotting the data with so.Bar() and so.Stack(). the contributions of the positive and negative values are overlapping, as the negative bar segments start where the positive end. As one can see, in the first bar we have a contribution of +1 and -2. The -2 bar segment starts at +1 and goes to -1. My wish would be to separate the positive from the negative segments, so that in this example the -2 segment ranges from 0 to -2.

import pandas as pd
import seaborn.objects as so
df = pd.DataFrame(
    {
        "node": ["node1", "node1", "node2", "node2"],
        "carrier": ["H2", "AC", "H2", "AC"],
        "value": [1, -2, 3, -4],
    }
)

plot = so.Plot(df, x="node", y="value", color="carrier")
plot = plot.add(so.Bar(), so.Stack())
plot

The produced figure is

Hope I could make the point clear

[mwaskom]

Don't you get the expected result here if you just remove so.Stack()?

[FabianHofmann]

Unfortunately no, for the previous example it seems to work fine, but when having multiple negatives and positives, they do not stack on top.

import pandas as pd
import seaborn.objects as so
df = pd.DataFrame(
    {
        "node": ["node1", "node1", "node1", "node2", "node2", "node2"],
        "carrier": ["H2", "AC", "DC", "H2", "AC", "DC"],
        "value": [1, -2, -3, 3, -4, 2],
    }
)

plot = so.Plot(df, x="node", y="value", color="carrier")
plot = plot.add(so.Bar())
plot

yields

whereas I would wish to have all negative segments added on top each on the negative side and all positives on the positive side.

[FabianHofmann]

@mwaskom I hope I could make myself clear. If not, please let me know :)

[mwaskom]

Maybe you could plot the positive and negative values in separate layers