Calculating the mix of values within a pandas groupby

Question

Within a database I have fields 'DATE', 'country', 'itemID', and 'price'. What I'm trying to do is calculate the mix of items (which sum to 100%) across price bin by date group by country.

The following code calculates the number of objects by country/price bin/date (which is aggregated to weekly frequency) and I'm happy with the results.

bins = [0, 5000, 10000, 20000, 30000, 40000, float('inf')]
labels = ['<5k', '10k', '20k', '30k', '40k', '>40k']
df.groupby(['country', pd.Grouper(key='DATE', freq='1W'), pd.cut(df['price'], bins=bins, labels=labels, include_lowest=True)]).agg({'itemID': 'nunique'}).unstack('DATE')

What I'd like to do is, instead of showing me the number of items in each category, I get the percentage in each price bin for that country on that date.

So, for example:

	DATE	2022-01-30	2022-02-06
country	price
--------	--------
USA	<5k	0%	1%
USA	10k	5%	19%
USA	20k	55%	25%
USA	30k	30%	55%
USA	40k	2%	0%
USA	>40k	8%	0%
JAP	<5k	0%	1%
JAP	10k	25%	0%
JAP	20k	45%	15%
JAP	30k	20%	35%
JAP	40k	2%	25%
JAP	>40k	8%	25%

I tried the below to initially calculate the number of unique itemIDs by country and date and then accessing this in the second part:

def f(d):
    return d['itemID'] / d['nuniques']

df['nuniques'] = df.groupby(['country', pd.Grouper(key='DATE', freq='1W')])['itemID'].transform('nunique')
df.groupby(['country', pd.Grouper(key='DATE', freq='1W'), pd.cut(df['price'], bins=bins, labels=labels, include_lowest=True)]).agg({'itemID': 'nunique', 'nuniques': 'min'}).apply(lambda x: f(x), axis=1).unstack('AS_OF_DATE')

Excluding the cut function, it shows 100% across each country (as expected).

However, adding the cut function causes the sum of the items across the bins to exceed 100% (by <1%, but different amounts for each date).

Why would this be, and how can it be calculated more properly? At a guess, the discrepancy is something to do with the fact that not every itemID has an entry for every date within the weekly window (e.g. if it goes out of stock).