pandas groupby count based on conditions

Question

I am trying to add a column to a dataframe that would give me a count of the type of payment returns that a customer has on their account.

This is what the dataframe looks like:

CustomerID       Return$     Payment Method
000010           10          Credit Card
000010           15          Credit Card
000011           10          Check
000011           15          Credit Card
000011           10          Credit Card

This is the expected outcome:

CustomerID     Return$   Payment Method   CC Return Count  Check Return Count
000010           10        Credit Card         2                  0
000010           15        Credit Card         2                  0
000011           10        Check               2                  1
000011           15        Credit Card         2                  1
000011           10        Credit Card         2                  1

This is the code that I have tried, but it only gives me a column with boolean values:

return_df['CC Boolean']= return_df.groupby(['CustomerID'])['Payment 
Method'].apply(lambda x: x=='Credit Card')

This other piece of code gives a total count of payment regardless of payment method:

return_df['Counter']= return_df.groupby('Customer ID')['Payment Method'].transform('count')

Answer 1

method_dict = df.groupby('CustomerID')['Payment Method'].value_counts().unstack().fillna(0).to_dict()

df['CC Return Count'] = df['CustomerID'].map(method_dict['Credit Card'])
df['Check Return Count'] = df['CustomerID'].map(method_dict['Check'])

Method dict looks like:

{'Check': {10: 0.0, 11: 1.0}, 'Credit Card': {10: 2.0, 11: 2.0}}

output:

df

>>>


CustomerID  Return$ Payment Method  CC Return Count Check Return Count
0   10      10      Credit Card     2.0             0.0
1   10      15      Credit Card     2.0             0.0
2   11      10      Check           2.0             1.0
3   11      15      Credit Card     2.0             1.0
4   11      10      Credit Card     2.0             1.0

Answer 2

You can use groupby on CustomerID and then count the number that of rows with 'Check' and 'Credit Card' for each customer. Using transform will preserve the original structure:

df['check'] = df.groupby('CustomerID')['Payment Method'].transform(lambda x: sum(x == 'Check'))
df['credit'] = df.groupby('CustomerID')['Payment Method'].transform(lambda x: sum(x == 'Credit Card'))

Output:

   CustomerID  Return$ Payment Method  check  credit
0          10       10    Credit Card      0       2
1          10       15    Credit Card      0       2
2          11       10          Check      1       2
3          11       15    Credit Card      1       2
4          11       10    Credit Card      1       2