基于索引的style.format

Question

You can specify a format for each column by using df.style.format() , however, i want this behavior but then index based instead of column based. I realise its a bit more tricky because a column has a specific datatype, and a row can be mixed.

Is there a workaround to get it anyway? The df.style.apply() method has the flexibility, but i don't think it supports number formatting, only (CSS) styling.

Some sample data:

import pandas as pd

df = pd.DataFrame([[150.00, 181.00, 186.00],
                   [  5.85,   3.73,   2.12]], 
                  index=['Foo', 'Bar'],
                  columns=list('ABC'))

If i transpose the Dataframe, is easy:

mapper = {'Foo': '{:.0f}',
          'Bar': '{:.1f}%'}   

df.T.style.format(mapper)

But i want this formatting without transposing, something like:

df.style.format(mapper, axis=1)

Answer 1

You may not need to use the Styler class for this if the target is to re-format row values. You can use that mapper dictionary to match the formats you want, through a map and apply combination by row. The following should be a decent start:

df.apply(lambda s: s.map(mapper.get(s.name).format), axis=1)

Thanks!