Selecting the data in specific columns of a DataFrame is performed by using the [] operator. This can be passed either as a single object, or a list of objects. These objects are then used to lookup columns either by zero-based location, or by matching the objects to the values in the columns index.

Passing a single integer, or a list of integers, to [] will have the DataFrame object attempt to perform a location based lookup of the columns. The following code retrieves the data in the second and third columns:

In [22]:
   # get first and second columns (1 and 2) by location
   sp500[[1, 2]].head()

Out[22]:
            Price  Book Value
   Symbol                    
   MMM     141.14      26.668
   ABT      39.60      15.573
   ABBV     53.95       2.954
   ACN      79.79       8.326
   ACE     102.91      86.897

Selecting columns by passing a list of values will result in another DataFrame, with data copied from the original DataFrame. This is true, even...