折叠 Pandas 数据框中的行
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/43514019/
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
Collapsing rows in a Pandas dataframe
提问by user4896331
I'm trying to collapse rows in a dataframe that contains a column of ID data and a number of columns that each hold a different string. It looks like groupby is the solution, but it seems to be slanted towards performing some numeric function on the group - I just want to keep the text. Here's what I've got...
我正在尝试折叠包含一列 ID 数据和多列（每列包含不同的字符串）的数据框中的行。看起来 groupby 是解决方案，但它似乎倾向于在组上执行一些数字功能 - 我只想保留文本。这是我所拥有的...
I have a dataframe of the form:
我有一个如下形式的数据框：
index ID apples pears oranges
0 101 oranges
1 134 apples
2 576 pears
3 837 apples
4 576 oranges
5 134 pears
The columns are clean: so the apples column will only ever have the text "apples" in it, or it will be blank".
列是干净的：所以苹果列中只会有文本“苹果”，或者它会是空白的。
Where there are multiple entries under the same ID (in this example, on IDs 134 & 576), I want to collapse the rows together to get this:
如果同一 ID 下有多个条目（在本例中，ID 为 134 和 576），我想将行折叠在一起以获得以下结果：
index ID apples pears oranges
0 101 oranges
1 134 apples pears
2 576 pears oranges
3 837 apples
I could do this by iterating over the rows, but it seems like a non-pandas solution. Is there a better way?
我可以通过迭代行来做到这一点，但这似乎是一个非Pandas解决方案。有没有更好的办法？
回答by jezrael
You can use groupbywith aggregation ''.join, sumor max:
您可以使用groupby聚合''.join，sum或max：
#if blank values are NaN first replace to ''
df = df.fillna('')
df = df.groupby('ID').agg(''.join)
print (df)
apples pears oranges
ID
101 oranges
134 apples pears
576 pears oranges
837 apples
Also works:
也有效：
df = df.fillna('')
df = df.groupby('ID').sum()
#alternatively max
#df = df.groupby('ID').max()
print (df)
apples pears oranges
ID
101 oranges
134 apples pears
576 pears oranges
837 apples
Also if need remove duplicates per group and per column add unique:
此外，如果需要删除每组和每列的重复项，请添加unique：
df = df.groupby('ID').agg(lambda x: ''.join(x.unique()))
回答by piRSquared
Assuming blanks are ''
假设空白是 ''
option 1pivot_table
选项1pivot_table
df.pivot_table(['apples', 'pears', 'oranges'], 'ID', aggfunc=''.join)
option 2sortand take last row as ''will be sorted first
选项 2sort并取最后一行，因为''将首先排序
def f(df):
return pd.DataFrame(np.sort(df.values, 0)[[-1]], [df.name], df.columns)
df.set_index(
'ID', append=True
).groupby(level='ID', group_keys=False).apply(f)
Both yield
两者产量
apples oranges pears
ID
101 oranges
134 apples pears
576 oranges pears
837 apples
