Python 更快地对熊猫数据框中的子组中的行进行排名
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/26720916/
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
Faster way to rank rows in subgroups in pandas dataframe
提问by captain ahab
I have a pandas data frame that has is composed of different subgroups.
我有一个由不同子组组成的熊猫数据框。
df = pd.DataFrame({
'id':[1, 2, 3, 4, 5, 6, 7, 8],
'group':['a', 'a', 'a', 'a', 'b', 'b', 'b', 'b'],
'value':[.01, .4, .2, .3, .11, .21, .4, .01]
})
I want to find the rank of each id in its group with say, lower values being better. In the example above, in group A, Id 1 would have a rank of 1, Id 2 would have a rank of 4. In group B, Id 5 would have a rank of 2, Id 8 would have a rank of 1 and so on.
我想找到每个 id 在其组中的排名，例如，值越低越好。在上面的例子中，在 A 组中，Id 1 的等级为 1，Id 2 的等级为 4。在 B 组中，Id 5 的等级为 2，Id 8 的等级为 1，依此类推在。
Right now I assess the ranks by:
现在我通过以下方式评估排名：
Sorting by value.
df.sort('value', ascending = True, inplace=True)Create a ranker function (it assumes variables already sorted)
def ranker(df): df['rank'] = np.arange(len(df)) + 1 return dfApply the ranker function on each group separately:
df = df.groupby(['group']).apply(ranker)
按值排序。
df.sort('value', ascending = True, inplace=True)创建一个排名函数（它假设变量已经排序）
def ranker(df): df['rank'] = np.arange(len(df)) + 1 return df分别对每个组应用 ranker 函数：
df = df.groupby(['group']).apply(ranker)
This process works but it is really slow when I run it on millions of rows of data. Does anyone have any ideas on how to make a faster ranker function.
这个过程有效，但是当我在数百万行数据上运行它时它真的很慢。有没有人对如何制作更快的排名功能有任何想法。
采纳答案by Jeff
rank is cythonized so should be very fast. And you can pass the same options as df.rank()hereare the docs for rank. As you can see, tie-breaks can be done in one of five different ways via the methodargument.
rank 是 cythonized 所以应该非常快。您可以传递与df.rank()此处相同的选项，因为rank. 如您所见，可以通过method参数以五种不同方式之一进行抢七。
Its also possible you simply want the .cumcount()of the group.
也有可能你只是想要.cumcount()组的。
In [12]: df.groupby('group')['value'].rank(ascending=False)
Out[12]:
0 4
1 1
2 3
3 2
4 3
5 2
6 1
7 4
dtype: float64
回答by Quentin Febvre
Working with a big DataFrame (13 million lines), the method rank with groupby maxed out my 8GB of RAM an it took a really long time. I found a workaround less greedy in memory , that I put here just in case:
使用大数据帧（1300 万行），使用 groupby 的方法排名最大化了我的 8GB RAM，这花了很长时间。我找到了一个不那么贪婪的解决方法，我把它放在这里以防万一：
df.sort_values('value')
tmp = df.groupby('group').size()
rank = tmp.map(range)
rank =[item for sublist in rank for item in sublist]
df['rank'] = rank
