Question

我需要计算某一多指数主要标题的多指数小标题的数目。

This question is related to my previous question here: read_csv: Multiple Header Rows. Single header row has duplicates within it, but no column is duplicated when both header rows considered If that link doesn t pass, the search term on stackoverflow would be: "read_csv: Multiple Header Rows. Single header row has duplicates within it, but no column is duplicated when both header rows considered"

采用同样的样本目录:

Sample csv file:

Exp,2023-09-22,2023-09-22,2023-09-29,2023-10-06,2023-10-13  
Code,A,B,A,A,A  
Item,,,,,  
260,1,8,17,31,42  
270,2,9,18,32,43  
280,3,10,19,33,44  
290,4,11,20,34,45  
300,5,12,21,35,46

And reading the csv into a data frame like this:


Gives Output:
print(df)

Exp  2023-09-22     2023-09-29 2023-10-06 2023-10-13
Code          A   B          A          A          A
Item                                                
270           2   9         18         32         43
280           3  10         19         33         44
290           4  11         20         34         45
300           5  12         21         35         46



Question:  How do I count the number of "Code" entries for a given Exp heading?
For example:
How many Code entries are listed for Exp  2023-09-22 ?  Answer: 2 (they are  A  and  B )
How many Code entries are listed for Exp  2023-10-13 ?  Answer: 1
令人非常赞赏的是,任何帮助掩盖了这一问题。
我曾尝试过各种镜头(df.columns[col_idx])和其他器具,但我没有看到任何东西。 关于可接受的解决办法,我要么用数字,如图2或图1,要么用我随后可以计算的姓名或价值清单(或无论它们是否认为)。 无论如何,这些法典要么是“A”要么是“B”,要么是“A”&”B”。

Answer 1

Code

单位+规模(按栏目0级)

df.groupby(level=0, axis=1).size()

产出:

Exp
2023-09-22      2
2023-09-29      1
2023-10-06      1
2023-10-13      1

如果贵国的法典清单不是独一无二的(例如,如果是A、A、B在2023-09-22下,如果你想回去2),则按法典使用。

df.droplevel(level=-1, axis=1).columns.to_frame(index=False).groupby( Exp )[ Code ].nunique()

Sample csv file:

And reading the csv into a data frame like this:

Gives Output:

友情链接