pandas 基于具有特定值的行创建一个新的数据框
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/51004029/
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
Create a new dataframe based on rows with a certain value
提问by Tom
I have a large dataframe of transactions which I want to break into two smaller dataframes based on a certain column ("Type"). If "Type" is "S" then add that entire row to the "cust_sell" dataframe, and if "Type" is "P" to the "cust_buy" dataframe. I am using a for loop, but this is only adding the index value to the dataframe. Any help is appreciated!
我有一个很大的交易数据框，我想根据某个列（“类型”）将其分成两个较小的数据框。如果“Type”为“S”，则将整行添加到“cust_sell”数据框中，如果“Type”为“P”，则添加到“cust_buy”数据框中。我正在使用 for 循环，但这只是将索引值添加到数据帧。任何帮助表示赞赏！
from win32com.shell import shell, shellcon
import pandas as pd
filename = (shell.SHGetFolderPath(0, shellcon.CSIDL_PERSONAL, None, 0)) + '\MSRB T-1_test.xlsx'
wb = pd.read_excel(filename, sheet_name='T1-20062017', index_col=0, header=0)
cust_buy = []
cust_sell = []
# Create a list of customer buys and sells separately
for i in wb.index:
if wb['Type'][i] == 'S':
cust_sell.append([i])
elif wb['Type'][i] == 'P':
cust_buy.append([i])
回答by Ankur Sinha
You do not need to write loops. You can do it easily with pandas.
您不需要编写循环。您可以使用Pandas轻松完成。
Assuming your dataframe looks like this:
假设您的数据框如下所示：
import pandas as pd
mainDf = pd.DataFrame()
mainDf['Type'] = ['S', 'S', 'S', 'P', 'P', 'S', 'P', 'S']
mainDf['Dummy'] = [1, 2, 3, 4, 5, 6, 7, 8]
To create dataframe for S and P types, you can just do this:
要为 S 和 P 类型创建数据框，您可以这样做：
cust_sell = mainDf[mainDf.Type == 'S']
cust_buy = mainDf[mainDf.Type == 'P']
cust_sell output:
cust_sell 输出：
Type Dummy
0 S 1
1 S 2
2 S 3
5 S 6
7 S 8
cust_buy output:
cust_buy 输出：
Type Dummy
3 P 4
4 P 5
6 P 7
回答by Seb
Like @trollster said, it is indeed better to create dataframes for cust_sell and cust_buy. But let's understand what is not working with your code. When you do:
就像@trollster 所说的那样，为 cust_sell 和 cust_buy 创建数据帧确实更好。但是让我们了解什么不适用于您的代码。当你这样做时：
for i in wb.index
it means i will take the values of wb.index. And when you print wb.index, you get:
这意味着我将采用 wb.index 的值。当你打印 wb.index 时，你会得到：
Int64Index([0, 1, 2, 3, 4, 5, 6, 7], dtype='int64')
Meaning i will take the values 1,2,3 ... So when you do:
这意味着我将取值 1,2,3 ... 所以当你这样做时：
cust_sell.append([i])
what you are doing is adding to the list_cust_sell a list with inside a single element, i (an integer). If you want to add the entire row, you should use:
您正在做的是向 list_cust_sell 添加一个包含单个元素 i （整数）的列表。如果要添加整行，则应使用：
cust_sell.append(list(wb.loc[i,:]))
You will end up with a list of lists, each one containing a row
您最终会得到一个列表列表，每个列表包含一行
回答by jpp
Using dict+ groupbyyou can create a dictionary of dataframes. This solution does not require you to manually specify all unique types and is more easily extendable than a manual loop.
使用dict+groupby您可以创建数据框字典。此解决方案不需要您手动指定所有唯一类型，并且比手动循环更易于扩展。
Data from @trollster.
来自@trollster 的数据。
res = dict(tuple(mainDf.groupby('Type')))
{'P': Type Dummy
3 P 4
4 P 5
6 P 7,
'S': Type Dummy
0 S 1
1 S 2
2 S 3
5 S 6
7 S 8}
