Python Pandas dataFrame - Columns selection

Question

I have a Pandas dataFrame object train_df with say a column called "ColA" and a column "ColB". It has been loaded from a csv file with columns header using read_csv

I obtain the same results when I code:

pd.crosstab(train_df['ColA'], train_df['ColB'])

or

pd.crosstab(train_df.ColA, train_df.ColB)

Is there any difference in these 2 ways of selecting columns?

When I request to print the type it's the same: pandas.core.series.Series

Answer 1

No difference

pd.crosstab(train_df['ColA'], train_df['ColB']) is recommended to prevent possible errors.

For example, if you have a column named count and if you type train_df.count it will give an error. train_df['count'] won't give an error.

Answer 2

If you only want to select a single column, there is no difference between the two ways.

However, the dot notation doesn't allow you to select multiple columns, whereas you can use dataframe[['col1', 'col2']] to select multiple columns (which returns a pandas.core.frame.DataFrame instead of a pandas.core.series.Series ).