是否可以在 Altair 中对具有多个数据源的分层图表进行分面？

Question

My question is similar to the one here , however I cannot figure out how to specify the data correctly.

I am trying to create a layered bar and tick chart with faceting. You cannot layer a faceted chart, so I have to specify the data and faceting in the parent layered chart. Following the example in the question linked, I have added a second data set for layering while also keeping the facet:

bars = pd.DataFrame({
    'year': np.repeat([2018, 2019], 6),
    'cat1': np.tile(['a', 'a', 'a', 'b', 'b', 'b'], 2),
    'cat2': np.tile(list('XYZ'), 4),
    'vals': np.arange(1, 13)
})

ticks = pd.DataFrame({
    'year': np.repeat([2018, 2019], 2),
    'cat1': np.tile(['a', 'b'], 2),
    'limit': [8, 13, 25, 32]
})

b = alt.Chart().mark_bar().encode(
    x='year:O',
    y='vals',
    color='cat2',
).properties(width=100)

t = alt.Chart(ticks).mark_tick(color='black', thickness=2, width=40).encode(
    x='year:O',
    y='limit',
)

alt.layer(b, t, data=bars).facet(
    'cat1:N',
)

My problem is that I would like the tick marks to be facetted as well. Is this possible without combining the datasets?

For example, this solution works by merging the data together first, but it is also drawing multiple ticks on top of each other:

bars_and_ticks = bars.merge(ticks, on=['year', 'cat1'])

t = alt.Chart().mark_tick(color='black', thickness=2, width=40).encode(
    x='year:O',
    y='limit',
)

alt.layer(b, t, data=bars_and_ticks).facet(
    'cat1:N',
)

Answer 1

You can only facet by a single dataset. I would approach what you want to do by combining your inputs into a single dataframe, and building the chart from there. For example:

import pandas as pd
import numpy as np
import altair as alt

bars = pd.DataFrame({
    'year': np.repeat([2018, 2019], 6),
    'cat1': np.tile(['a', 'a', 'a', 'b', 'b', 'b'], 2),
    'cat2': np.tile(list('XYZ'), 4),
    'vals': np.arange(1, 13)
})

ticks = pd.DataFrame({
    'year': np.repeat([2018, 2019], 2),
    'cat1': np.tile(['a', 'b'], 2),
    'limit': [8, 13, 25, 32]
})

data = pd.merge(bars, ticks, how='left', on=['year', 'cat1'])

b = alt.Chart(data).mark_bar().encode(
    x='year:O',
    y='vals',
    color='cat2',
).properties(width=100)

t = alt.Chart(data).mark_tick(color='black', thickness=2, width=40).encode(
    x='year:O',
    y='max(limit):Q',
)

alt.layer(b, t).facet(
    'cat1:N',
)