Customized float formatting in a pandas DataFrame

Question

I have a DataFrame:

   0       1
0  3.000   5.600
1  1.200   3.456

and for presentation purposes I would like it to be converted to

   0    1
0  3    5.6
1  1.2  3.456

What is the elegant way to achieve this (without looping inefficiently over entries of the DataFrame)?

Or perhaps more generally: is there a way to set pandas up such that it is always doing this? E.g. one of the pandas options?

Notice that pd.options.display.float_format = '{:,.0f}'.format will not work, as it would give a fixed number of decimals, rather than having it vary across entries of the DataFrame as I indicated above.

Answer 1

In [188]: df
Out[188]:
       a      b       c
0 1.0000 2.2460  2.0000
1 3.0000 4.4920  6.0000
2 5.0000 6.7380 10.0000

In [189]: pd.options.display.float_format = '{:,.2f}'.format

In [190]: df.apply(lambda x: x.astype(int) if np.allclose(x, x.astype(int)) else x)
Out[190]:
   a    b   c
0  1 2.25   2
1  3 4.49   6
2  5 6.74  10

UPDATE:

In [222]: df
Out[222]:
       0      1
0 3.0000 5.6000
1 1.2000 3.4560

In [223]: df.applymap(lambda x: str(int(x)) if abs(x - int(x)) < 1e-6 else str(round(x,2)))
Out[223]:
     0     1
0    3   5.6
1  1.2  3.46

NOTE: be aware that .applymap() method is pretty slow as it's doing map(func, series) for each series in the DataFrame