Get indexes of multiple pandas columns by names

Question

I want to get the numerical indexes of a selection of pandas dataframe columns.

With one column it's very simple:

nonzero(df.columns.values == 'conditionA')

but with multiple elements? I have something that works but is verbose and hugly:

df = pd.DataFrame(columns=['conditionF', 'conditionB', 'conditionA', 'conditionD', 'conditionC'])

cols_to_find = ['conditionA', 'conditionB', 'conditionC']
[i for i in range(len(df.columns.values)) if df.columns.tolist()[i] in cols_to_find ]

Better ideas?

That's the answer! It was just there... I was looking for a isin() function and did not go for a method! Tx! — Gioelelm, Apr 08 '14 at 10:52
@behzad.nouri please post it as the answer so Gioelelm would be able to accept it and show the others that the issue is clearly solved — Bruno Gelb, Apr 08 '14 at 10:55
@behzad.nouri: nice, but it doesn't keep the order from the list of column names, it rearranges them in increasing order. — smci, Dec 14 '17 at 20:50
Gioelelm, I posted a working solution two years ago. You are able to click accept on answers. — smci, Oct 21 '19 at 21:47

smci · Answer 1 · 2021-08-09T21:57:23.437

11

This works, and also preserves order:

[df.columns.get_loc(col) for col in cols_to_find]
[2, 1, 4]

List comprehensions are your friend.

edited Aug 09 '21 at 21:57

answered Dec 19 '17 at 03:14

smci

32,567
20
113
146

@MTKnife: ok but AFAIK it works fine for the question which doesn't mention multiindex, can you post an example (with unique column names) where `get_loc` returns multiple values? Then we can see about handling that case. – smci Dec 20 '19 at 08:15
Actually, my apologies: I just noticed you were looking at columns, and with columns, you're unlikely to have duplicates. It would be a problem with identically named rows (including, say, multiple rows with the same 0-level value in a `MultiIndex`). – MTKnife Dec 20 '19 at 16:31

score 0 · Answer 2 · answered Mar 06 '22 at 12:20

df=pd.DataFrame(columns=['conditionF', 'conditionB', 'conditionA', 'conditionD', 'conditionC'])
def search():
    search=['conditionA', 'conditionB', 'conditionC']
    c=len(search)
    for i in df.columns.values:
        print(i)
        if i in search:
            c-=1
    return c==0
search()

Get indexes of multiple pandas columns by names

2 Answers2

Linked