Why does .loc assignment with two sets of brackets result in NaN in a pandas.DataFrame?

Question

I have a DataFrame:

	name	age
0	Paul	25
1	John	27
2	Bill	23

I know that if I enter:

df[['name']] = df[['age']]

I'll get the following:

	name	age
0	25	25
1	27	27
2	23	23

But I expect the same result from the command:

df.loc[:, ['name']] = df.loc[:, ['age']]

But instead, I get this:

	name	age
0	NaN	25
1	NaN	27
2	NaN	23

For some reason, if I omit those square brackets [] around column names, I'll get exactly what I expected. That is the command:

df.loc[:, 'name'] = df.loc[:, 'age']

gives the right result:

	name	age
0	25	25
1	27	27
2	23	23

Why does two pairs of brackets with .loc result in NaN? Is it some sort of a bug or is it intended behaviour? I can't figure out the reason for such a behaviour.

Answer 1

That's because for the loc assignment all index axes are aligned, including the columns: Since age and name do not match, there is no data to assign, hence the NaNs.

You can make it work by renaming the columns:

df.loc[:, ["name"]] = df.loc[:, ["age"]].rename(columns={"age": "name"})

or by accessing the numpy array:

df.loc[:, ["name"]] = df.loc[:, ["age"]].values