How can I style Pandas multiindex dataframe columns using threshold conditioning?

Question

I got one multiindex dataframe as below.

I would like to highlight all the "diff" columns, with different background colors on negative/positive values.

e.g. something like

if '-' in diff.value:
'background-color: lightgreen'
else:
'background-color: pink'

How do I write a styler for it?

I see another similar issue in pandas multiindex style highlight a row

But my dataframe seems not able to groupby as the recommendation in that thread.

Also the method in that thread does not contain threshold conditioning

Answer 1

You can use subset to highlight the row which contains -

def highlight_minus(row):
    return np.where(row.astype(str).str.contains('-'),
                    'background-color: lightgreen',
                    'background-color: pink')

idx = pd.IndexSlice
slice_ = idx[:, idx[:,'diff']]
s = df.style.apply(highlight_minus, axis=1, subset=slice_)
s.to_html('76374687.html')