Reference Column with spaces through the “dot” method / attribute access (df.column_name)

Question

This is a very basic question, but I believe it’ll help someone out there (including me).

When a data frame is created and we want to reference the column name, we can do so by the following:

df[“Column Name Example”]

However, we can also reference a column name by the ‘dot’ method, such as this example:

df.column_name_example

The second option works because of the underscores within the column name. What if the column name has spaces? How do we use the ‘dot’ method to reference column names?

Answer 1

This is not possible, and not due to pandas design but rather to python's syntax.

You can only use valid python names as attribute.

• it must start with a letter or underscore
• it can only contain letters, 0-9, and _

other limitations

Other than this limitation, this method of indexing is highly unrecommended, as it is not programmatic and can lead to bugs if the name clashes with a pandas method.

Example:

df = pd.DataFrame({'sum': [1,2,3]})

df.sum + 1

Output:

TypeError: unsupported operand type(s) for +: 'method' and 'int'

In contrast:

df['sum'] + 1

0    2
1    3
2    4
Name: sum, dtype: int64

In summary, you should really avoid using attribute access, only doing it for quick testing if you want to spare 3 keystrokes...