将Excel列的层次结构转换为行的层次结构

问题描述 投票:0回答:1

我有3列的Excel。

OrgunitCode    ParentOrgunitCode     OrgunitName

1500                  Nan                 Head_Org

3200                  1500              2nd_level_Org

3201                  1500              other_2nd_lever_Org

..............................................................

971248197              827484           n_level_Org

103048197              513834           n2_level_Org

我需要把它转换为完整的层次结构行(用Excel或Python)。

OrgunitName OrgunitCode    OrgunitName(lvl_2) OrgunitCode(lvl_2) .. OrgunitName(lvl_5) OrgunitCode(lvl_5)

Head_Org        1500        2nd_level_Org          3200                  n_level_Org         971248197 

Head_Org        1500        2nd_level_Org          3200                  n2_level_Org         513834

..............................................................

所以我想使用PivotTables,或在excel中查找,但如果我从 "Parent "开始搜索,我不知道有多少 "Childrens",所以我很困惑。而且我不知道如何从最低点开始,因为如何找到他们。

python excel pandas tree hierarchy
1个回答
0
投票

哦......使用层次结构的乐趣。总是很有趣。

首先,在处理父子关系时,我总是建议建立一个组织节点,比如''代表第1级;'1'、'2'......'n'代表第2级;'11'、'21'、'22'......。'nn'代表第3级等等。这将使你的层次结构工作变得更容易(如排序、寻找不同部门的后代、兄弟姐妹和祖先等)。

在你的情况下,你并不真的需要一个组织节点,但我已经把它包括在D列中供将来参考。

假设第1行是个头,数据的组织形式是:

A: OrgunitCode 
B: ParentOrgunitCode 
C: OrgunitName

然后在第2行添加这些公式。

D: OrgNode (not used but useful in future)
=IF(COUNTIFS($A:$A,B2)=0,"/"&A2&"/",VLOOKUP(B2,$A:$D,4,FALSE)&$A2&"/")

E: Hierarchy level
=IF(COUNTIFS($A:$A,$B2)=0,1,VLOOKUP($B2,$A:$E,5,FALSE)+1)

接下来是你要求的层次结构。将这个公式分别粘贴到单元格E2和F2中,拖动它们覆盖所有相关行和层次结构即可。

F: OrgunitName
=IF($E2=(COLUMN()-4)/2,$C2,IF($E2<(COLUMN()-4)/2,"",VLOOKUP($B2,$A:F,COLUMN(),FALSE)))

G: OrgunitCode
=IF($E2=(COLUMN()-5)/2,$A2,IF($E2<(COLUMN()-5)/2,"",VLOOKUP($B2,$A:G,COLUMN(),FALSE)))

需要注意的是,G和F中的查找列是相对的,避免在向右扩展公式时,改变数字或增加数字行来参考。但是,如果列的分割不完全如上,可能要用数字或数字引用来代替'COLUMN()'部分。

还有一点:VLOOKUP在这里也可以用,但要注意的是,对于Excel来说,这是一个很繁重的操作,所以如果你有一个非常大的层次结构,你可以考虑使用SQL中的层次结构函数、MySQL中的递归函数或类似的函数。

公式剖析

EDIT:按要求进行公式解释(我觉得很乱,但不知道有什么更好的办法)。

D: OrgNode (not used but useful in future)
=IF(COUNTIFS($A:$A,B2)=0,"/"&A2&"/",VLOOKUP(B2,$A:$D,4,FALSE)&$A2&"/")

Builds a hierarchy node with the parent/child relationship seperated by '/':
'COUNTIFS($A:$A,B2)=0'              <- checks to see if there is a parent. 
"/"&A2&"/"                          <- ...If no, then this is the/a top node, so take the OrgCode and wrap it in '/'
VLOOKUP(B2,$A:$D,4,FALSE)&$A2&"/")  <- Else, look up the parent's orgnode and add this department's OrgCode and '/' to the parent's.


E: Hierarchy level
=IF(COUNTIFS($A:$A,$B2)=0,1,VLOOKUP($B2,$A:$E,5,FALSE)+1)

Just determines the hierarchy level of the dept in question.
=IF(COUNTIFS($A:$A,$B2)=0       <- checks to see if there is a parent.
1                               <- if not, then make this hierarchy level 1
VLOOKUP($B2,$A:$E,5,FALSE)+1)   <- Else, lookup the hierarchy level of the parent and add 1

F: OrgunitName
=IF($E2=(COLUMN()-4)/2,$C2,IF($E2<(COLUMN()-4)/2,"",VLOOKUP($B2,$A:F,COLUMN(),FALSE)))

=IF($E2=(COLUMN()-4)/2,     <- checks to see if the current hierarchy column is equal to the hierarchy level of the current department 
$C2                         <- if so, return the dept name
IF($E2<(COLUMN()-4)/2       <- else check if the current hierarchy column is lower than the hierarchy level of the current department
""                          <- if so, remain blank
VLOOKUP($B2,$A:F,COLUMN(),FALSE))) <- else lookup the value in this column of the parent department.

G: OrgunitCode 
=IF($E2=(COLUMN()-5)/2,$A2,IF($E2<(COLUMN()-5)/2,"",VLOOKUP($B2,$A:G,COLUMN(),FALSE)))

- Same explanation as above, just a different column reference and returns department code instead
=IF($E2=(COLUMN()-5)/2
$A2
IF($E2<(COLUMN()-5)/2
""
VLOOKUP($B2,$A:G,COLUMN(),FALSE)))

并将其刻画出来。

(COLUMN()-4)/2  <- Instead of hardcoding the hierarhy levels, this uses the column index to calculate the relevant hierarchy level in the given column. Column F is index 6: (6-4)/2 = 1. Column H is index 8: (8-4)/2 = 2. etc.

COLUMN()        <- In the VLOOKUP reference: We all know VLOOKUP is annoying beacause we have to manually change the number in the col_index_num parameter when dragging across columns. But similar to above, we can use the column index number as a relative reference. Hence her, column F is index 6 and will look up the 6th column. Column G will return the 7th, H the 8th... you get the point.
© www.soinside.com 2019 - 2024. All rights reserved.