CODEWITHSUNDEEP
pythonpandasnumpy
mokiliii Lo
mokiliii Lo

Reputation: 637

Pandas - re-arrange columns by ascending order

I have the following dataframe:

+------+-----+-----+-----+-----+
| Name | 1   | 4   | 3   | 2   |
+------+-----+-----+-----+-----+
| ...  | ... | ... | ... | ... |
+------+-----+-----+-----+-----+

And I need my dataframe to be re-arranged like this:

+------+-----+-----+-----+-----+
| Name | 1   | 2   | 3   | 4   |
+------+-----+-----+-----+-----+
| ...  | ... | ... | ... | ... |
+------+-----+-----+-----+-----+

with text first, then numbers in increasing order.

I have tried doing:

df[sorted(df.columns)]

But it doesn't work as the columns are of string type.

Upvotes: 1

Views: 158

Answers (1)

jezrael
jezrael

Reputation: 862511

Use DataFrame.sort_index with convert columns to integers with to_numeric with errors='coerce', so missing values for non numbers are created. Then is used parameter na_position='first' for starts by this columns and then are values sorted by numerics:

df = pd.DataFrame(columns=['Name','1','10','7', '6'])

df=df.sort_index(axis=1,key=lambda x: pd.to_numeric(x,errors='coerce'),na_position='first')
print (df)
Empty DataFrame
Columns: [Name, 1, 6, 7, 10]
Index: []

If values are only numbers is used Index.astype:

df = pd.DataFrame(columns=['1','10','7', '6'])

df = df.sort_index(axis=1, key=lambda x: x.astype(int))
print (df)
Empty DataFrame
Columns: [1, 6, 7, 10]
Index: []

Or:

df = df[sorted(df.columns,key=lambda x:int(x))]
print (df)
Empty DataFrame
Columns: [1, 6, 7, 10]
Index: []

Upvotes: 4

Related Questions