Look at the default presentation of a dataframe¶

Let's load in a standard csv file and print it out using the defaults. Then we will see how we can use Panda's styling to make it easier to extract information. The dataset we are using are the campaign contributions for the 2016 presidential elections. The candidates have been sorted by the total amount of money the raised.

Throughout this, we will look at the changes that styling makes just to the first few rows of our dataframe.

Adjust the floats (all)¶

We see that we have a mix of scientific notation (2.311e+08), as well as the large numbers in the Transfers column that are very difficult to read. We can ask for all floats from a dataframe to be formatted as

• floating point (no exponents),
• with two decimal places,
• and , separators between the hundreds, thousands, millions, etc

Let's see this in action:

This is now the default for all dataframes in this workbook, not just this one. This is because we have told Pandas to adjust its settings globally.

Overriding the percentages and dollars¶

We can override the global again, to put a $ sign in front of all the floats, as follows:

Note that the Percent individual also got a dollar sign. This isn't what we wanted! We can set the format column by column if we want, using the .style method:

Note that we have lost the dollar signs, and we have scientific notation back. That is because doing .style no longer returns a dataframe (instead it returns a Styler object), and the Pandas defaults only apply to dataframes. In particular, if we try running a dataframe method, such as .head() on the styler, it won't work:

We can check that the Pandas defaults persist by looking at a dataframe again (instead of a Styler object):

We see that the dollar signs are back (as well as, sadly, on the percentage column).

If we use a format dictionary, we have to set the format on each column we want, or accept the defaults. Luckily, we have select_dtypes, which selects all columns of the type we are interested in. We can generate the dictionary with the default we want, and then override specific values.

Here is this trick in action:

Display surnames in caps (without changing underlying data)¶

We can also change the style of string columns. For example, this is how we would change the display of the surnames:

Note the surnames themselves haven't changed:

Set background color¶

We can even do things like change the background color depending on the value of the cell. The syntax for this is

def some_func(value):
    """Function should return valid CSS as a string
    e.g. the return string might be
    '''background-color: <some CSS color name>;
       color: <some CSS color name>;
       font-size: <some number>pt;
    '''
    """
    ...

Here we will implement a function, party_color, the will return a CSS string that changes the color of the cell based on the value of the string (red for Republican, blue for Democrats, green for the Greens, and grey for everyone else).

This function is then passed to applymap as follows: