Feature request: add an optional values parameter to Styler.background_gradient

[dclong]

Sometimes we want to apply background_gradient to some columns according to values in another column. There's no easy way to do this currently. It can be made possible by adding an optional values parameter. And when the values parameter is specified, it is used to generate background colors.

[TomAugspurger]

Can you provide an example?

[simonjayhawkins]

if I understand correctly what your trying to achieve, you could render the HTML directly yourself, using df.to_html(escape=False)

something like this?..

import pandas as pd
from IPython.display import HTML, display

from . import datasets


def make_clickable(val):
    return '<a target="_blank" href="https://www.amazon.co.uk/dp/{}">{}</a>'.format(val, val)


def image_html(row):
    if type(row['SmallImage_URL']) != float:
        return '<img src="{}" style="height: {}px; width: {}px;">' \
            .format(row['SmallImage_URL'], int(row['SmallImage_Height']), int(row['SmallImage_Width']))
    else:
        return ''


def add_image_to_dataframe(df):
    df = df.copy()
    columns = df.columns.tolist()
    image_metadata_columns = ['SmallImage_Height',
                              'SmallImage_Width', 'SmallImage_URL']
    df = datasets.add_item_attributes(df, image_metadata_columns)
    df['image'] = df.apply(image_html, axis=1)
    df = df.drop(columns=image_metadata_columns)
    return df[['image'] + columns]


def render(df, add_image=True):
    if not df.empty is True:
        df = df.copy().fillna("")
        if 'asin' not in df.columns:
            df = datasets.add_asin(df)
        if (add_image):
            df = add_image_to_dataframe(df)
        df['asin'] = df.asin.apply(make_clickable)
        with pd.option_context('display.max_colwidth', -1):
            html = df.to_html(index=False, escape=False)
        display(HTML(html))

[dclong]

For example, I want to have background gradient colors for 2 columns Ctrl and Test in a DataFrame, but the background gradient colors are decided according to the difference of the 2 columns. I achieved it as below.

def bg_grad(s, m, M, cmap='PuBu', low=0, high=0):
    rng = M - m
    norm = colors.Normalize(m - (rng * low), M + (rng * high))
    normed = norm(s.values)
    c = [colors.rgb2hex(x) for x in plt.cm.get_cmap(cmap)(normed)]
    return ['background-color: %s' % color for color in c]


lift = (df.Test - df.Ctrl) / df.Ctrl 
df.style.apply(lambda _: bg_grad(lift, 
    m=-lift.max(),
    M=lift.max(),
    cmap='RdYlGn',
))

[TomAugspurger]

It's not clear to me what the API would look like. IIRC, there's nothing else in Styler where a property for a column is computed by the values in another array.

It is unfortunate that you needed to duplicate the code in Styler._background_gradient with you bg_grad function.

[attack68]

I think this is fairly easy to deal with with an introduction of a new argument. gmap or gradient map just overrides the data in the input series with the specifically provided gmap. That way you can style a whole dataframe along an axis using only one row (or column), and achieve the OP's intent. eg.

df.style.background_gradient(cmap='PuBu', gmap=df['col1'], axis=0)

I'll get around to this at some point..