是否可以根据类别之一的值对Altair分组条形图的列进行排序? [英] Is it possible to sort the columns of an Altair grouped bar chart based on the value of one of the categories?

问题描述

我有以下图表-

我希望能够对列进行排序(不是单个组的单个条-我已经知道如何做到)，即订购3个子图-如果可以的话-我选择的任何类别( a ， b 或 c )的值.

我尝试使用 alt.SortField 和 alt.EncodeSortField ，它们在图表上略有移动，但是如果您更改类别以查看是否他们实际上在工作.

代码-

 将altair导入为alt将熊猫作为pd导入dummy = pd.DataFrame({'place':['Asia'，'Antarctica'，'Africa'，'Antarctica'，'Asia'，'Africa'，'Africa'，'Antarctica'，'Asia']，'类别':['a'，'a'，'a'，'b'，'b'，'b'，'c'，'c'，'c']，'值':[5,2，3,4,3,5,6,9,5]})alt.Chart(dummy).mark_bar().encode(x = alt.X('category')，y ='value'，column = alt.Column('place:N'，sort = alt.SortField(field ='value'，order ='descending'))，color ='category'，) 

我知道 alt.Column('place:N'，sort = alt.SortField(field ='value'，order ='descending'))似乎不正确，因为我没有针对任何类别，所以我也尝试了 x = alt.X('category'，sort = alt.SortField(field ='c'，order ='descending'))，但是它也不起作用.

预期产量(假设降序)-

• 如果我想按'c'排序，则应该首先在中间一列，然后是左列，最后是右列.
• 它似乎已经按'b'排序了.
• 如果我想按'a'排序，则应首先选择右列，然后是左列，最后是中间列.

解决方案

这有点涉及，但是您可以通过一系列转换来做到这一点:

• a

  然后按"a" 进行排序:

    alt.Chart(dummy).transform_calculate(key ="datum.category =='a'").transform_joinaggregate(sort_key ="argmax(key)"，groupby = ['place']).transform_calculate(sort_val ='datum.sort_key.value').mark_bar().encode(x = alt.X('category')，y ='value'，column = alt.Column('place:N'，sort = alt.SortField("sort_val"，order ="descending"))，color ='category'，) 

  I have the following chart -
  
  I'd like to be able to sort the columns (NOT the individual bars of a single group - I know how to do that already), i.e order the 3 sub-chart - if you will - based on the value of any category(a,b or c) I choose.

  I tried using alt.SortField and alt.EncodeSortField, they move around the charts a bit, but don't actually work if you change the category to see if they actually work.

  Code -

  import altair as alt
  import pandas as pd
  
  dummy = pd.DataFrame({'place':['Asia', 'Antarctica','Africa', 'Antarctica', 'Asia', 'Africa', 'Africa','Antarctica', 'Asia'],'category':['a','a','a','b','b','b','c','c','c'],'value':[5,2,3,4,3,5,6,9,5]})
  alt.Chart(dummy).mark_bar().encode(
      x=alt.X('category'),
      y='value',
      column=alt.Column('place:N', sort=alt.SortField(field='value', order='descending')),
      color='category',
  )
  

  I know that alt.Column('place:N', sort=alt.SortField(field='value', order='descending')), doesn't seem correct, since I am not targeting any category, so I tried x=alt.X('category', sort=alt.SortField(field='c', order='descending')), too, but it doesn't work either.

  Expected Output (assuming descending order)-

  • If I want to order by 'c', then middle column should be first, followed by left and finally right column.
  • It already seems ordered by 'b'.
  • If I want to order by 'a', then right column should be first, followed by left and finally middle column.

  解决方案

  This is a bit involved, but you can do this with a series of transforms:

  • a Calculate Transform to select the value you want to sort on
  • a Join-Aggregate Transform with argmax to join the desired values to each group
  • another calculate transform to pull-out the specific field within this result that you would like to sort by

  It looks like this, first sorting by "c":

  import altair as alt
  import pandas as pd
  
  dummy = pd.DataFrame({'place':['Asia', 'Antarctica','Africa', 'Antarctica', 'Asia', 'Africa', 'Africa','Antarctica', 'Asia'],'category':['a','a','a','b','b','b','c','c','c'],'value':[5,2,3,4,3,5,6,9,5]})
  alt.Chart(dummy).transform_calculate(
      key="datum.category == 'c'"
  ).transform_joinaggregate(
      sort_key="argmax(key)", groupby=['place']
  ).transform_calculate(
      sort_val='datum.sort_key.value'  
  ).mark_bar().encode(
      x=alt.X('category'),
      y='value',
      column=alt.Column('place:N', sort=alt.SortField("sort_val", order="descending")),
      color='category',
  )
  

  Then sorting by "a":

  alt.Chart(dummy).transform_calculate(
      key="datum.category == 'a'"
  ).transform_joinaggregate(
      sort_key="argmax(key)", groupby=['place']
  ).transform_calculate(
      sort_val='datum.sort_key.value'  
  ).mark_bar().encode(
      x=alt.X('category'),
      y='value',
      column=alt.Column('place:N', sort=alt.SortField("sort_val", order="descending")),
      color='category',
  )
  

  这篇关于是否可以根据类别之一的值对Altair分组条形图的列进行排序?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！