Finding a cumulative sum of unique values in a pandas dataframe column

Question

I'm struggling to find the cumulative sum of the income value by day in this example df. I would like to see the cumulative value for each day of the week add up, so Monday + Monday. Tuesday + Tuesday. etc.

Example df:

df = pd.DataFrame({'day': ['Mon','Tue','Wed', 'Mon', 'Tue', 'Wed'],
                  'date': ['2002-01-02', '2002-01-03', '2002-01-04', '2002-01-08', '2002-01-09', '2002-01-10'],
                  'income': [40, 60, 40, 100, 55, 32]
                 })

I would like to add a cumsum() column to this df, that only adds the income for each day.

My attempt that just adds every row cumulatively in a table of plot:

df['cum_income_by_day'] = df['income'].cumsum() 
Monday = df[df['day']] == 'Monday'
sn.lineplot(data = Monday, x="Date", y="cum_income_by_day")

Attempt 2 throws a 'setting with copy' warning that is valid, my results aren't accurate - not sure what is happening but i can see the cumulation of the first few values is wrong.

Monday = df[df['day'] == 'Monday']
df['cum_income_by_day'] = Monday['income'].cumsum()

I thought maybe the answer is in groupby given i want to do this for every day, not just Mondays, but i just get one cumulative value. I tried a for loop (I'm new, i'm still learning) and couldn't crack it. Any advice much appreciated.

Ideal output looks like this:

If you like my question upvote it so i have enough points to upvote your answers.

Answer 1

You have to groupby first!

>>> df['cumsum_income'] = df.groupby('day')['income'].cumsum()

---

   day        date  income  cumsum_income
0  Mon  2002-01-02      40             40
1  Tue  2002-01-03      60             60
2  Wed  2002-01-04      40             40
3  Mon  2002-01-08     100            140
4  Tue  2002-01-09      55            115
5  Wed  2002-01-10      32             72