python - 在带有标记文本的 altair 中对条形图进行排序时出现问题

Question

chart_df= alt.Chart(df).mark_bar().encode(
    x = 'value',
    y = alt.Y('name', sort='-x'),
    color = 'variable'
)

为了将每个条形图的值添加为文本，我使用了以下代码，但我丢失了排序的条形图。

chart_df_text = chart_df.mark_text().encode(
    x = 'text_margin_from_bar:Q',
    text = 'human_readable_value:Q',
).transform_calculate(
    human_readable_value = expr.toString(expr.floor(datum.value/10**7)),
    text_margin_from_bar = datum.value + (datum.value/expr.abs(datum.value))*1000000000
    # i have negetive and positive numbers, so for have a space between number and bar, i do this
)

添加

y = alt.Y('name', sort='-x'),

到 chart_df_text 但我仍然有问题。我读了另一个有我问题的问题，说问题是 altair 的版本，但我在最后一个。

Answer 1

javascript 控制台中的警告告诉您为什么这不起作用:

[Warning] Dropping sort property {"field":"value","op":"sum"} as unioned domains only support boolean or op "count", "min", and "max".
[Warning] Domains that should be unioned has conflicting sort properties. Sort will be set to true.

您可以通过在您的排序中使用受支持的 op 来解决这个问题。例如:

import pandas as pd
import altair as alt
from altair import expr, datum

df = pd.DataFrame({
    'value': [-3E9, -4E9, 6E9, 1E10, -8E9],
    'name': ['Bob', 'Sue', 'Tim', 'Ann', 'Bill'],
    'variable': range(5)
})

chart_df= alt.Chart(df).mark_bar().encode(
    x = 'value',
    y = alt.Y('name', sort=alt.EncodingSortField('value', op='min', order='descending')),
    color = 'variable'
)

chart_df_text = chart_df.mark_text().encode(
    x = 'text_margin_from_bar:Q',
    text = 'human_readable_value:Q',
).transform_calculate(
    human_readable_value = expr.toString(expr.floor(datum.value/10**7)),
    text_margin_from_bar = datum.value + (datum.value/expr.abs(datum.value))*1000000000
)

chart_df + chart_df_text