Format certain floating dataframe columns into percentage in pandas

I am trying to write a paper in IPython notebook, but encountered some issues with display format. Say I have following dataframe df, is there any way to format var1 and var2 into 2 digit decimals and var3 into percentages.

       var1	       var2         var3	
id												
0	 1.458315	 1.500092	-0.005709	
1	 1.576704	 1.608445	-0.005122	 
2	 1.629253	 1.652577	-0.004754	 
3	 1.669331	 1.685456	-0.003525	
4	 1.705139	 1.712096	-0.003134	
5	 1.740447	 1.741961	-0.001223	
6	 1.775980	 1.770801	-0.001723	 
7	 1.812037	 1.799327	-0.002013	 
8	 1.853130	 1.822982	-0.001396	 
9	 1.943985	 1.868401	 0.005732

The numbers inside are not multiplied by 100, e.g. -0.0057=-0.57%.

The accepted answer suggests to modify the raw data for presentation purposes, something you generally do not want. Imagine you need to make further analyses with these columns and you need the precision you lost with rounding.

You can modify the formatting of individual columns in data frames, in your case:

output = df.to_string(formatters={
    'var1': '{:,.2f}'.format,
    'var2': '{:,.2f}'.format,
    'var3': '{:,.2%}'.format
})
print(output)

For your information '{:,.2%}'.format(0.214) yields 21.40%, so no need for multiplying by 100.

You don't have a nice HTML table anymore but a text representation. If you need to stay with HTML use the to_html function instead.

from IPython.core.display import display, HTML
output = df.to_html(formatters={
    'var1': '{:,.2f}'.format,
    'var2': '{:,.2f}'.format,
    'var3': '{:,.2%}'.format
})
display(HTML(output))

Update

As of pandas 0.17.1, life got easier and we can get a beautiful html table right away:

df.style.format({
    'var1': '{:,.2f}'.format,
    'var2': '{:,.2f}'.format,
    'var3': '{:,.2%}'.format,
})

You could also set the default format for float :

pd.options.display.float_format = '{:.2%}'.format

Use '{:.2%}' instead of '{:.2f}%' - The former converts 0.41 to 41.00% (correctly), the latter to 0.41% (incorrectly)

replace the values using the round function, and format the string representation of the percentage numbers:

df['var2'] = pd.Series([round(val, 2) for val in df['var2']], index = df.index)
df['var3'] = pd.Series(["{0:.2f}%".format(val * 100) for val in df['var3']], index = df.index)

The round function rounds a floating point number to the number of decimal places provided as second argument to the function.

String formatting allows you to represent the numbers as you wish. You can change the number of decimal places shown by changing the number before the f.

p.s. I was not sure if your 'percentage' numbers had already been multiplied by 100. If they have then clearly you will want to change the number of decimals displayed, and remove the hundred multiplication.

Often times we are interested in calculating the full significant digits, but for the visual aesthetics, we may want to see only few decimal point when we display the dataframe.

In jupyter-notebook, pandas can utilize the html formatting taking advantage of the method called style.

For the case of just seeing two significant digits of some columns, we can use this code snippet:

Given dataframe

import numpy as np
import pandas as pd

df = pd.DataFrame({'var1': [1.458315, 1.576704, 1.629253, 1.6693310000000001, 1.705139, 1.740447, 1.77598, 1.812037, 1.85313, 1.9439849999999999],
          'var2': [1.500092, 1.6084450000000001, 1.652577, 1.685456, 1.7120959999999998, 1.741961, 1.7708009999999998, 1.7993270000000001, 1.8229819999999999, 1.8684009999999998],
          'var3': [-0.0057090000000000005, -0.005122, -0.0047539999999999995, -0.003525, -0.003134, -0.0012230000000000001, -0.0017230000000000001, -0.002013, -0.001396, 0.005732]})

print(df)
       var1      var2      var3
0  1.458315  1.500092 -0.005709
1  1.576704  1.608445 -0.005122
2  1.629253  1.652577 -0.004754
3  1.669331  1.685456 -0.003525
4  1.705139  1.712096 -0.003134
5  1.740447  1.741961 -0.001223
6  1.775980  1.770801 -0.001723
7  1.812037  1.799327 -0.002013
8  1.853130  1.822982 -0.001396
9  1.943985  1.868401  0.005732

Style to get required format

    df.style.format({'var1': "{:.2f}",'var2': "{:.2f}",'var3': "{:.2%}"})

Gives:

     var1	var2	var3
id			
0	1.46	1.50	-0.57%
1	1.58	1.61	-0.51%
2	1.63	1.65	-0.48%
3	1.67	1.69	-0.35%
4	1.71	1.71	-0.31%
5	1.74	1.74	-0.12%
6	1.78	1.77	-0.17%
7	1.81	1.80	-0.20%
8	1.85	1.82	-0.14%
9	1.94	1.87	0.57%

Update

If display command is not found try following:

from IPython.display import display

df_style = df.style.format({'var1': "{:.2f}",'var2': "{:.2f}",'var3': "{:.2%}"})

display(df_style)

Requirements

• To use display command, you need to have installed Ipython in your machine.
• The display command does not work in online python interpreter which do not have IPyton installed such as https://repl.it/languages/python3
• The display command works in jupyter-notebook, jupyter-lab, Google-colab, kaggle-kernels, IBM-watson,Mode-Analytics and many other platforms out of the box, you do not even have to import display from IPython.display

As suggested by @linqu you should not change your data for presentation. Since pandas 0.17.1, (conditional) formatting was made easier. Quoting the documentation:

> You can apply conditional formatting, the visual styling of a DataFrame depending on the data within, by using the DataFrame.style property. This is a property that returns a pandas.Styler object, which has useful methods for formatting and displaying DataFrames.

For your example, that would be (the usual table will show up in Jupyter):

df.style.format({
    'var1': '{:,.2f}'.format,
    'var2': '{:,.2f}'.format,
    'var3': '{:,.2%}'.format,
})

Just another way of doing it should you require to do it over a larger range of columns

using applymap

df[['var1','var2']] = df[['var1','var2']].applymap("{0:.2f}".format)
df['var3'] = df['var3'].applymap(lambda x: "{0:.2f}%".format(x*100))

applymap is useful if you need to apply the function over multiple columns; it's essentially an abbreviation of the below for this specific example:

df[['var1','var2']].apply(lambda x: map(lambda x:'{:.2f}%'.format(x),x),axis=1)

Great explanation below of apply, map applymap:

https://stackoverflow.com/questions/19798153/difference-between-map-applymap-and-apply-methods-in-pandas

As a similar approach to the accepted answer that might be considered a bit more readable, elegant, and general (YMMV), you can leverage the map method:

# OP example
df['var3'].map(lambda n: '{:,.2%}'.format(n))

# also works on a series
series_example.map(lambda n: '{:,.2%}'.format(n))

Performance-wise, this is pretty close (marginally slower) than the OP solution.

As an aside, if you do choose to go the pd.options.display.float_format route, consider using a context manager to handle state per this parallel numpy example.

style.format is vectorized, so we can simply apply it to the entire df (or just its numerical columns):

df[num_cols].style.format('{:,.3f}')