When a DataFrame operates directly with one of the arithmetic or comparison operators, each value of each column gets the operation applied to it. Typically, when an operator is used with a DataFrame, the columns are either all numeric or all object (usually strings). If the DataFrame does not contain homogeneous data, then the operation is likely to fail. Let's see an example of this failure with the college dataset, which contains both numeric and object data types. Attempting to add 5 to each value of the DataFrame raises a TypeError as integers cannot be added to strings:

>>> college = pd.read_csv('data/college.csv')
>>> college + 5
TypeError: Could not operate 5 with block values must be str, not int

To successfully use an operator with a DataFrame, first select homogeneous data. For this recipe, we will select all the columns that begin with UGDS_. These columns represent the fraction of undergraduate students by race. To get started, we import the data and use the institution name as the label for our index, and then select the columns we desire with the filter method:

>>> college = pd.read_csv('data/college.csv', index_col='INSTNM')
>>> college_ugds_ = college.filter(like='UGDS_')
>>> college_ugds_.head()

This recipe uses multiple operators with a DataFrame to round the undergraduate columns to the nearest hundredth. We will then see how this result is equivalent to the round method.