在altair-viz中对连接的图像进行分面

如何解决在altair-viz中对连接的图像进行分面

这是对Altair: Sorting faceted “text” chart not reflecting expectation的后续查询。我用另一列修改了数据框。我的目标是对“标记”（Marker）列中的对象进行分组，以便最终得到一个多面的图形，每个图形都有条形图和文本。数据框如下。在尝试进行构面之前，我先粘贴单个图像的外观。随后，我将粘贴代码和我的最佳尝试。我想知道altair代码中是否有办法做到这一点，或者您是否建议在altair代码外使用for循环进行groupby。

,Bug,UnkNown,Level,LDA_score,p_value,Marker
0,a,4.10808792666,Low,3.43193376894,0.0381678194757,GM
1,b,2.80231776318,High,2.86568860404,0.048078814719199996,GM
2,c,1.55012602444,3.0159901714,0.047006554908300004,GM
3,d,2.11298173821,2.94493334678,0.0120363750248,GM
4,e,2.08807237447,2.9096371889,0.0149437560986,GM
5,f,2.762619332479999,2.52323422148,0.040652301139,GM
6,g,4.390454714340001,3.85075499081,0.029978515680400004,GM
7,h,3.32306083381,3.01988462626,0.0244409015043,GM
8,i,2.84614167157,2.97142565384,0.0438396924694,GM
9,j,4.51419624602,3.84190054285,0.0460224914387,GM
10,k,4.027450677669999,3.52319882849,0.0113390729281,IFN
11,l,4.26967903787,3.8458771734,0.00548234585386,IFN
12,m,1.7823168924,2.50020069082,0.0203578926278,IFN

此代码和图像如下所示，而没有尝试按最后一列进行分组：

y_sort = alt.EncodingSortField(field='LDA_score',order='descending')

bars = alt.Chart(df).mark_bar().encode(
    alt.X('LDA_score',title='LDA_score (log10)',axis=alt.Axis(titleFontSize=14)),alt.Y("Bug:N",sort=y_sort,axis=alt.Axis(title=None,labelFontStyle='italic')),color=alt.Color('Level:N',legend=alt.Legend(title=None,labelFontSize=12,orient='right')),# scale=alt.Scale(domain=['>12weeks','<12weeks'],range=['green','red'])),row=alt.Row('Level:N',header=alt.Header(title=None,labelFontSize=0),spacing=0),).resolve_scale(
    y='independent'
)

text = alt.Chart(df).mark_text().encode(
    alt.Text('p_value:Q',format='.3e'),axis=None),).resolve_scale(
    y='independent'
).properties(width=50,title="p_value"
)

FinalChart = alt.hconcat(bars,text,spacing=-10)\
    .configure_title(anchor='end',fontStyle='italic',fontSize=14)\
    .configure_axis(grid=True,gridOpacity=0.5).configure_view(opacity=0.5)
FinalChart.display()

我随后尝试在altair / vega中使用facet选项。我要粘贴代码，然后粘贴：

y_sort = alt.EncodingSortField(field='LDA_score',).resolve_scale(
    y='independent'
).facet(column='Marker')

text = alt.Chart(df).mark_text().encode(
    alt.Text('p_value:Q',title="p_value"
).facet(column='Marker')

FinalChart = alt.hconcat(bars,gridOpacity=0.5).configure_view(opacity=0.5)
FinalChart.display()

我认为这可能很直观，但是我想要的是这样的东西：

解决方法

您追求的是刻面连接图表；不幸的是，Altair或Vega-Lite不支持此功能。但是您可以通过手工构造构面来解决此问题：构面操作基本上是一个过滤器和一个concat，因此您可以像这样构造所需的图表：

import altair as alt
import pandas as pd
from io import StringIO

df = pd.read_csv(StringIO("""\,Bug,Unknown,Level,LDA_Score,p_value,Marker
0,a,4.10808792666,Low,3.43193376894,0.0381678194757,GM
1,b,2.80231776318,High,2.86568860404,0.048078814719199996,GM
2,c,1.55012602444,3.0159901714,0.047006554908300004,GM
3,d,2.11298173821,2.94493334678,0.0120363750248,GM
4,e,2.08807237447,2.9096371889,0.0149437560986,GM
5,f,2.762619332479999,2.52323422148,0.040652301139,GM
6,g,4.390454714340001,3.85075499081,0.029978515680400004,GM
7,h,3.32306083381,3.01988462626,0.0244409015043,GM
8,i,2.84614167157,2.97142565384,0.0438396924694,GM
9,j,4.51419624602,3.84190054285,0.0460224914387,GM
10,k,4.027450677669999,3.52319882849,0.0113390729281,IFN
11,l,4.26967903787,3.8458771734,0.00548234585386,IFN
12,m,1.7823168924,2.50020069082,0.0203578926278,IFN
"""))

y_sort = alt.EncodingSortField(field='LDA_Score',order='descending')

bars = alt.Chart(df).mark_bar().encode(
    alt.X('LDA_Score',title='LDA_Score (log10)',axis=alt.Axis(titleFontSize=14)),alt.Y("Bug:N",sort=y_sort,axis=alt.Axis(title=None,labelFontStyle='italic')),color=alt.Color('Level:N',legend=alt.Legend(title=None,labelFontSize=12,orient='right')),# scale=alt.Scale(domain=['>12weeks','<12weeks'],range=['green','red'])),row=alt.Row('Level:N',header=alt.Header(title=None,labelFontSize=0),spacing=0),column=alt.Column('Marker:N',title=None),).resolve_scale(
    y='independent'
)

text = alt.Chart(df).mark_text().encode(
    alt.Text('p_value:Q',format='.3e'),axis=None),).resolve_scale(
    y='independent'
).properties(width=50,title="p_value"
)

FinalChart = alt.hconcat(
    bars.transform_filter('datum.Marker == "GM"'),text.transform_filter('datum.Marker == "GM"'),bars.transform_filter('datum.Marker == "IFN"'),text.transform_filter('datum.Marker == "IFN"')
).configure_title(anchor='end',fontStyle='italic',fontSize=14)\
 .configure_axis(grid=True,gridOpacity=0.5).configure_view(opacity=0.5)
FinalChart.display()