Pandas：排序数据透视表
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/10595327/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
Pandas: Sort pivot table
提问by mrmagooey
Just trying out pandas for the first time, and I am trying to sort a pivot table first by an index, then by the values in a series.
只是第一次尝试熊猫，我试图首先按索引对数据透视表进行排序，然后按系列中的值排序。
So far I've tried:
到目前为止，我已经尝试过：
table = pivot_table(sheet1, values='Value', rows=['A','B'], aggfunc=np.sum)
# Sorts by value ascending, can't change to descending
table.copy().sort()
table
# The following gives me the correct ordering in values, but ignores index
sorted_table = table.order(ascending=False)
sorted_table
# The following brings me back to the original ordering
sorted_table = table.order(ascending=False)
sorted_table2 = sorted_table.sortlevel(0)
sorted_table2
What's the correct way to sort a pivot table by index then value?
按索引然后按值对数据透视表进行排序的正确方法是什么？
回答by Wes McKinney
Here is a solution that may do what you want:
这是一个可以满足您要求的解决方案：
key1 = table.index.labels[0]
key2 = table.rank(ascending=False)
# sort by key1, then key2
sorter = np.lexsort((key2, key1))
sorted_table = table.take(sorter)
The result would look like this:
结果如下所示：
In [22]: table
Out[22]:
A B
bar one 0.698202
three 0.801326
two -0.205257
foo one -0.963747
three 0.120621
two 0.189623
Name: C
In [23]: table.take(sorter)
Out[23]:
A B
bar three 0.801326
one 0.698202
two -0.205257
foo two 0.189623
three 0.120621
one -0.963747
Name: C
This would be good to build into pandas as an API method. Not sure what it should look like though.
将其作为 API 方法构建到 Pandas 中会很好。不知道它应该是什么样子。
