Moving Pandas Dataframe Columns to the start

Question

I've got a pandas dataframe with almost a thousand columnds

The column titles are like

[smth,smth,smth,smth.....a,b,c,d,e]

how would I re arrange the columns to move A,B,C,D,E to the start:

[a,b,c,d,e,smth,smth......]

do you want to reverse the order of your columns? like [```df.iloc[:, ::-1]```](https://stackoverflow.com/questions/27817718/reverse-dataframe-column-order)? — sophocles, Dec 26 '21 at 17:26
what do you mean by `[smth, smth, smth, ...]`? Are they all the same except for the last few? — Pierre D, Dec 26 '21 at 17:32
I don't want to reverse them I just want them in the order with the last five becoming the first 5. (the last becoming the 5th etc) — Adi Krish, Dec 26 '21 at 20:32
Does this answer your question? [Move column by name to front of table in pandas](https://stackoverflow.com/questions/25122099/move-column-by-name-to-front-of-table-in-pandas) — E_net4 - Krabbe mit Hüten, Jan 03 '22 at 12:13
See also: https://stackoverflow.com/questions/13148429/how-to-change-the-order-of-dataframe-columns — E_net4 - Krabbe mit Hüten, Jan 03 '22 at 12:13

score 0 · Answer 1 · answered Dec 26 '21 at 17:21

If I were you I would just use the included .pop() method that is built in to pandas.

So in your case I would do something like this: You will end up with a dataFrame where the column the pop method was used on is now the first and it will subsequently shift all the rest.

first_column = df.pop('A')

You could continue to do this for each of the other columns and it would work well, and if you have so much data that it becomes cumbersome to do it this way you could just run a loop.

There is also some good info from pandas on this:

https://www.geeksforgeeks.org/how-to-move-a-column-to-first-position-in-pandas-dataframe/

score 0 · Answer 2 · answered Dec 26 '21 at 17:21

A clean and efficient way is to use reindex:

cols = list(df.columns)
df.reindex(columns=cols[-5:]+cols[:-5])

Example:

df = pd.DataFrame([], columns=['X', 'Y', 'Z', 'A', 'B', 'C', 'D', 'E'])
print(df)

cols = list(df.columns)
df = df.reindex(columns=cols[-5:]+cols[:-5])
print(df)

output:

Empty DataFrame
Columns: [X, Y, Z, A, B, C, D, E]
Index: []

Empty DataFrame
Columns: [A, B, C, D, E, X, Y, Z]
Index: []

score 0 · Accepted Answer · answered Dec 26 '21 at 17:36

if you simply want to get the last n columns to move to first .You could change the columns order and over write the data frame with that selection

import pandas as pd
import numpy as np
df = pd.DataFrame(np.random.randint(0,100,size=(100, 5)), columns=list('ABCDE'))
cols = list(df.columns)
cols_shift = 2
new_cols = []
new_cols.extend(cols[-cols_shift:])
new_cols.extend(cols[:-cols_shift])
new_cols

df = df[new_cols]
df

score 0 · Answer 4 · answered Dec 26 '21 at 17:36

An explicit way is as follow. Note that it doesn't even matter where the columns that we want first actually are.

# example setup
cols = 'foo,bar,smth,smth_else,a,b,c,d,e'.split(',')
df = pd.DataFrame(np.random.randint(0,10, size=(4, len(cols))), columns=cols)

Then, say the columns you want first are ['a', 'b', 'c', 'd', 'e']:

first = ['a', 'b', 'c', 'd', 'e']

out = df[first + [k for k in df.columns if k not in first]]

# or:
out = df[pd.Index(first).append(df.columns.difference(first))]


>>> out
   a  b  c  d  e  foo  bar  smth  smth_else
0  7  1  0  2  5    7    2     1          9
1  6  7  7  4  3    7    3     1          2
2  8  9  3  6  2    0    1     5          8
3  1  2  0  3  3    2    4     1          4

Moving Pandas Dataframe Columns to the start

4 Answers4