How can I replace NaN values with other values in the CustomerID column if they have a same InvoiceNo?

Question

InvoiceNo	StockCode	Quantity	InvoiceDate	UnitPrice	CustomerID	Country
573415	23434	20	2011-10-30 16:17:00	0.79	13607.0	UK
:--------	:---------	:---------	:-------------------	:---------	:----------	:-------
539050	22480	2	2010-12-15 16:21:00	1.25	12577.0	France
:--------	:---------	:---------	:-------------------	:---------	:----------	:-------
573415	23434	1	2011-01-13 14:29:00	4.13	NaN	UK
:--------	:---------	:---------	:-------------------	:---------	:----------	:-------
573151	84692	25	2011-10-27 20:09:00	0.42	17602.0	UK
:--------	:---------	:---------	:-------------------	:---------	:----------	:-------
539050	22480	4	2011-12-02 16:39:00	10.79	NaN	UK
:--------	:---------	:---------	:-------------------	:---------	:----------	:-------

Here I would like to replace the NaN values in the CustomerID with the values in CustomerID column if the InvoiceNo values are same.

Answer 1

Assuming that each invoice id has only 1 unique customer id something like this should work:

df["CustomerID"] = df.groupby("InvoiceNo")["Customer ID"].apply(lambda x: x.ffill().bfill())

The code above simply makes groups based on the customer id and forward an backward fills all NA values.