CODEWITHSUNDEEP
pythonpandaspandas-styles
Autumn
Autumn

Reputation: 3756

Default float format for Pandas styling

I have a large number of dataframes to output in Jupyter. The columns are a mix of strings, ints, and floats. The floats need mostly to be '%.2f', but a small subset require specific formatting -- mostly percentages and float-as-int.

The specific formatting is easy. But setting a default float format for Styling doesn't seem to exist.

The only option I've found so far is to write a function that applies custom formats to some float columns, and the default format to all the others, while ignoring the strings and ints.

It seems awfully kludgy to have to explicitly write the same format to 90% of my columns. Is there a way to set a same-as-except default style by dtype?

Upvotes: 2

Views: 2971

Answers (2)

JMKõ
JMKõ

Reputation: 81

For pandas, I find the most useful solution is to include the following line early on in the script:
pd.options.display.float_format = "{:,.0f}".format
Of course, the number of decimal digits as well as the "thousands" delimiter can be adjusted to meet your needs.
Health warning - By using this, all output (e.g., print, display, describe, etc.) will have the float output similarly adjusted. This can complicate problem-solving of your script as the numbers displayed may not be what is displayed.

Upvotes: 0

Ivan Sudos
Ivan Sudos

Reputation: 1483

Currently there is no such option for Styler (it is still under development so you can propose the idea). And as far as I get having Styler is requred for you. If so and if you need Styler only for views you can set new styling variable with default all-floats formatted: 

float_cols = [c for c in df.dtypes.index  if 'float' in str(df.dtypes[c])]
s = df.style.format(dict(zip(float_cols,  [lambda x: "{:.2f}".format(x)]*10)))

Having this done you can use variable s for any further styling manipulation like:

s = s.applymap(...)
s = s.format(...)
etc...

You can also try to subclass pandas dataframe to have you own dataframe class with the predefined style as in the code above: Pandas subclassing guide

Upvotes: 3

Related Questions