6

Question

Is there a way to format only a specific dataframe?

I've seen examples of formatting specific columns of a single dataframe (Example 1) or set the entire pandas library to a default option (Example 2). However, I haven't seen an option for formatting a specific dataframe without explicitly listing each column.

Setup

import pandas as pd
import numpy as np

# Setup first example
data = np.random.random((3,4))
df = pd.DataFrame(data)
print df
#          0         1         2         3
#0  0.384326  0.364187  0.084034  0.012376
#1  0.114784  0.298068  0.087634  0.828207
#2  0.255923  0.438617  0.820652  0.266964

Example 1 - Change format for specific column(s) in a single dataframe

df[3] = df[3].map('${:,.2f}'.format)
print df
#          0         1         2      3
#0  0.384326  0.364187  0.084034  $0.01
#1  0.114784  0.298068  0.087634  $0.83
#2  0.255923  0.438617  0.820652  $0.27

Example 2 - Change format for all pandas dataframes (including new ones)

pd.options.display.float_format = '${:,.2f}'.format
print(df)
#      0     1     2      3
#0 $0.38 $0.36 $0.08  $0.01
#1 $0.11 $0.30 $0.09  $0.83
#2 $0.26 $0.44 $0.82  $0.27

data2 = np.random.random((4,3))
df2 = pd.DataFrame(data2)
print df2
#      0     1     2
#0 $0.60 $0.37 $0.86
#1 $0.28 $0.06 $0.97
#2 $0.19 $0.68 $0.99
#3 $0.06 $0.88 $0.82

I was looking for an option like example 2, except it won't apply the formatting to future dataframes. Thanks!

EDIT - My apologies, I should've been clearer about the formatting. Example 1 changes the data type while Example 2 doesn't. I was hoping to not have to convert between data types (if possible). E.g. The first example changes from floats to non-null objects:

df.info()
#<class 'pandas.core.frame.DataFrame'>
#Int64Index: 3 entries, 0 to 2
#Data columns (total 4 columns):
#0    3 non-null float64
#1    3 non-null float64
#2    3 non-null float64
#3    3 non-null object
#dtypes: float64(3), object(1)
Will
  • 11,276
  • 9
  • 68
  • 76

2 Answers2

5

I think your best bet is to pass a formatter to to_string

In [283]: print df.to_string(float_format='${:,.2f}'.format)
      0     1     2     3
0 $0.53 $0.01 $0.75 $0.61
1 $0.54 $0.33 $0.42 $0.47
2 $0.28 $0.67 $0.71 $0.53

Although that won't stay with the dataframe. You could do some kind of monkey-patch like this.

In [286]: from functools import partial

In [287]: df.to_string = partial(df.to_string, float_format='${:,.2f}'.format)

In [288]: print df
      0     1     2     3
0 $0.53 $0.01 $0.75 $0.61
1 $0.54 $0.33 $0.42 $0.47
2 $0.28 $0.67 $0.71 $0.53
chrisb
  • 49,833
  • 8
  • 70
  • 70
  • Thanks chrisb! I upvoted, but went and accepted the other answer since it didn't require an import. I really like your use of partial. – Will Aug 19 '14 at 14:59
4

How about just loop through the columns like this:

for i in range(len(df.columns)):
    df[i] = df[i].map('${:,.2f}'.format)

or just:

df.applymap('${:,.2f}'.format)
lessthanl0l
  • 1,035
  • 2
  • 12
  • 21