How to add new column to pandas group? Pandas forgets the column

Question

I have a pandas dataframe indexed by createdAt grouped by pid (participant identifiers). The created at column are unix timestamps. Now I would like to add a new column to each group with a string representation of the day and month so so that I can group again by day.

But pandas seems to forget about the new column I have added?

So I have this:

def to_daymonth(timestamp: int):
    datime_obj = datetime.fromtimestamp(timestamp)
    return datime_obj.strftime('%d %b')

for pid, group in bypid:
    group['date'] = group.index.map(to_daymonth)
    print(group.date)  # Inside the for loop this prints the new column like 01 May etc.

# But outside of the for loop
 
print(bypid.get_group('12')['date']) # KeyError: 'date'
print(bypid.get_group('12').date) # AttributeError: 'DataFrame' object has no attribute 'date'

It seems to me that pandas is forgetting I added the date column? What am I missing here?

This is what I would like to do after it remembers the date column.

for pid, group in bypid:
    plt.figure()
    plt.title(pid)
    plt.plot((0,1), (0.5, 0.5))  # Lines for x and y in the middle
    plt.plot((0.5, 0.5), (0, 1))
    for date, dategroup in group.groupby('date'):
        plt.scatter(dategroup.euro, dategroup.dollar, label=date)
    plt.legend(loc='best')
    plt.show()