How to format heatmap annotation floats to only show the decimal component

Question

I have floats and I'm interested in just the decimal places of them because the integer part is not important (and always will be the same in my problem).

I have to use string formatting because the number will be displayed on seaborn plots. I'm using '0.2f' to display just two decimal, however I don't know how to remove the integer part.

Example:

I want to show .25 instead of 0.25

Details:

The implementation must work to format the values in seaborn.heatmap:

sns.heatmap(table, annot = True, fmt = '.2f') 

but I want to display just the decimals as described above.

Answer 1

• As per Why doesn't f-string formatting work for Pandas?, an f-string can be used in a list-comprehension with .apply, to format all of the columns in the dataframe, which can then be passed to annot=, with fmt=''.
  • Each value is formatted with f'{v:.2f}'[-3:] where .2f rounds to two decimal places, and [-3:] slices the last three characters (e.g. .xx).
  • A typical pandas solution tries to avoid this usage of .apply, but given the size of a dataframe passed to sns.heatmap, this will not impact performance.

import seaborn as sns
import numpy as np  # for random data

# sample data
np.random.seed(20231118)
df = pd.DataFrame(np.random.random(size=(6, 6)))

# create a new f-string formatted dataframe; these will be str dtype
annot = df.apply(lambda col: [f'{v:.2f}'[-3:] for v in col], axis=0)

# plot with annotations
ax = sns.heatmap(data=df, fmt='', annot=annot, cbar=False)

df

          0         1         2         3         4         5
0  0.321406  0.647402  0.173481  0.885630  0.206138  0.584212
1  0.617528  0.782675  0.071267  0.858961  0.689791  0.093260
2  0.161312  0.069866  0.047737  0.804141  0.523107  0.635975
3  0.508007  0.890942  0.435791  0.281811  0.560090  0.384583
4  0.264682  0.501962  0.492275  0.143144  0.629373  0.445461
5  0.883968  0.207125  0.946935  0.222533  0.001788  0.622386

annot

     0    1    2    3    4    5
0  .32  .65  .17  .89  .21  .58
1  .62  .78  .07  .86  .69  .09
2  .16  .07  .05  .80  .52  .64
3  .51  .89  .44  .28  .56  .38
4  .26  .50  .49  .14  .63  .45
5  .88  .21  .95  .22  .00  .62

Answer 2

You can try nb -= int(nb) or decimalPart = nb - int(nb) to make the number be only the decimal part.

Answer 3

The following code extracts the 2 decimal points from a given floating point number.

float_number = 28.31
formatted_number = "{:.2f}".format(float_number).split('.')[1]

You can pass the above to a formatter as in here and get the plotting done per your requirement.