如何根据动态列值 select 行？

Question

首先，我有以下数据框 df_A

sector	SALES	EBIT	DPS
IT	xxxx	yyyy	zzz
ENERGY	xxxx	yyyy	zzz
FINANCE	xxxx	yyyy	zzz
CONSUMER	xxxx	yyyy	zzz

和另一个数据框df_B

NAME	sector	SALES	EBIT	DPS
AAPL	IT	xxxx	yyyy	zzz
BP	ENERGY	xxxx	yyyy	zzz
TGT	CONSUMER	xxxx	yyyy	zzz
MSFT	IT	xxxx	yyyy	zzz
HSBC	FINANCE	xxxx	yyyy	zzz
GOOG	IT	xxxx	yyyy	zzz
WMT	CONSUMER	xxxx	yyyy	zzz
META	IT	xxxx	yyyy	zzz
CVX	ENERGY	xxxx	yyyy	zzz
JPM	FINANCE	xxxx	yyyy	zzz
MCD	CONSUMER	xxxx	yyyy	zzz

等等

这只是一个例子，我有比这更大的数据框

我想做的是通过区分 df_B 的扇区来创建新的数据帧；

其中新创建的数据帧遵循 df_A["sectors"]

的顺序

最后将它们合并在一起，希望是水平格式

所以最后我希望我的输出看起来像

NAME	sector	SALES	EBIT	DPS	NAME	sector	SALES	EBIT	DPS	NAME	sector	SALES	EBIT	DPS	NAME	sector	SALES	EBIT	DPS
AAPL	IT	xxxx	yyyy	zzz	BP	ENERGY	xxxx	yyyy	zzz	HSBC	FINANCE	xxxx	yyyy	zzz	WMT	CONSUMER	xxxx	yyyy	zzz
MSFT	IT	xxxx	yyyy	zzz	CVX	ENERGY	xxxx	yyyy	zzz	JPM	FINANCE	xxxx	yyyy	zzz	TGT	CONSUMER	xxxx	yyyy	zzz
GOOG	IT	xxxx	yyyy	zzz	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	MCD	CONSUMER	xxxx	yyyy	zzz
META	IT	xxxx	yyyy	zzz

如果上面的横排不行，竖排table也行

我是 python 的菜鸟，我尝试使用 for 循环、字典、loc/iloc，但不知何故 none 我的代码工作正常...

非常感谢任何帮助

Answer 1

创建 N 个数据帧，每个扇区一个，然后将它们连接成一个：

out = pd.concat([pd.DataFrame(df_B[df_B['sector'] == sector].to_dict('records'))
                    for sector in df_A['sector'].unique().tolist()], axis=1)
print(out)

# Output
   NAME sector SALES  EBIT  DPS NAME  sector SALES  EBIT  DPS  NAME   sector SALES  EBIT  DPS NAME    sector SALES  EBIT  DPS
0  AAPL     IT  xxxx  yyyy  zzz   BP  ENERGY  xxxx  yyyy  zzz  HSBC  FINANCE  xxxx  yyyy  zzz  TGT  CONSUMER  xxxx  yyyy  zzz
1  MSFT     IT  xxxx  yyyy  zzz  CVX  ENERGY  xxxx  yyyy  zzz   JPM  FINANCE  xxxx  yyyy  zzz  WMT  CONSUMER  xxxx  yyyy  zzz
2  GOOG     IT  xxxx  yyyy  zzz  NaN     NaN   NaN   NaN  NaN   NaN      NaN   NaN   NaN  NaN  MCD  CONSUMER  xxxx  yyyy  zzz
3  META     IT  xxxx  yyyy  zzz  NaN     NaN   NaN   NaN  NaN   NaN      NaN   NaN   NaN  NaN  NaN       NaN   NaN   NaN  NaN

如何根据动态列值 select 行？

How to select rows based on dynamic column value?

python

loops

lines-of-code

pandas