How to sort Pandas df columns' values based on its position not labels?

Question

There must be an obvious answer but I could neither find it in the sort_values()'s docs nor in related questions posts*

by in df.sort_values() accept columns labels, but How do I sort using the columns' position ?

I came up with this cumbersome code

df.iloc[df.iloc[:,1].sort_values().index]

to sort on the second column.

I was thinking that something like df.sort_values(by=1,axis=1) would exist. So what is syntacticly simple and correct way to do that ?

*Related questions: 1, 2, 3, 4, 5, 5, 6

I'm not sure to understand. df.iloc[:, np.argsort(df.columns.values)] will reorder the columns in the df, right ? What I need is to sort the df on, for example, the 33th column's values. — Malik Koné
– Malik Koné, Commented Nov 5, 2017 at 6:34

jezrael · Accepted Answer · 2017-11-05 06:27:26Z

4

Simpliest is select columns names by position:

df = df.sort_values(df.columns[1])

answered Nov 5, 2017 at 6:27

jezrael

868k103 gold badges1.4k silver badges1.3k bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

yes ! I see. No shorter hand then ? Why not allowing positional int as a by value ?

In my opinion it is not implemented yet.