How to format row or column as percent

[mcp292]

Problem

How do you format the title of row or column facets as percent?

For normal axes it's alt.Y().axis(format="%"), but alt.Row().axis(format="%") does not work, reporting AttributeError: 'Row' object has no attribute 'axis'.

MWE

import altair as alt
import pandas as pd


source = pd.DataFrame(
    [
        {"a": "a1", "b": "b2", "c": "z", "p": "0.55"},
        {"a": "a1", "b": "b3", "c": "y", "p": "0.58"},
        {"a": "a2", "b": "b3", "c": "x", "p": "0.74"},
        {"a": "a2", "b": "b3", "c": "y", "p": "0.64"},
    ]
)

alt.Chart(source, width=60, height=alt.Step(8)).mark_bar().encode(
    alt.Y("c:N").axis(None),
    alt.X("a:N").title("Factor A"),
    alt.Color("c:N").title("settings").legend(orient="bottom", titleOrient="left"),
    alt.Row("p:Q"),
    # alt.Row("p:Q").axis(format="%"),
    alt.Column("b:N").title("Factor B"),
).save("test.png")

This graph makes no sense; I reached for the quickest example to modify.

[mcp292]

Bump.

[joelostblom]

You can use header instead of axis. Note that it seems like you also need to round the percentage so something like this would work alt.Row("p:Q").header(format=".0%"):

[mcp292]

Thank you so much!

This works but still some weird behavior I'm not fully understanding.

1. Why does format need a precision specifier? With alt.X().axis(format="%") this is not needed. Why does the header format behavior differ from the axis format behavior when the documentation doesn't?
2. Why do I get the following warning: WARN row encoding should be discrete (ordinal / nominal / binned).?

[joelostblom]

1. I am not sure, that was suprising to me too. I guess there are different default formatting rules between axes and headers.
2. That is because Vega-Lite recommends to use one of those types for the row encoding. You can specify it with p:N (and still do the formatting). I don't think anything will fail with quantitative, but not sure

[mcp292]

Thank you!