Pandas group by subtraction on aggregation

Question

I have a pandas dataframe df that has entries for account such that Person Name, Account id have credit and debit entries, for example

date        Name      transaction-type  tran
2013-03-05  john Doe   credit          10
2013-05-05  john Doe   debit           20
2012-06-01  jane Doe   credit          50

I wanted to group the transactions by date, name and transaction-type and aggregate the tran?. How could I do this? I was hoping to be able to do a reduce(numpy.subtract) on the tran column but I am not really sure on the correct syntax for Pandas.

Answer 1

IIUC, you simply want .groupby and then .sum() :

>>> df
                 date      Name transaction-type  tran
0 2013-03-05 00:00:00  john Doe           credit    10
1 2013-05-05 00:00:00  john Doe            debit    20
2 2012-06-01 00:00:00  jane Doe           credit    50
3 2012-06-01 00:00:00  jane Doe           credit    22
4 2012-06-02 00:00:00  jane Doe           credit    75
>>> df.groupby(["date", "Name", "transaction-type"]).sum()
                                      tran
date       Name     transaction-type      
2012-06-01 jane Doe credit              72
2012-06-02 jane Doe credit              75
2013-03-05 john Doe credit              10
2013-05-05 john Doe debit               20

See the section on groupby aggregation in the docs.

If you want the total signed value, you could get that too:

>>> df["tran"][df["transaction-type"] == "debit"] *= -1
>>> df.groupby(["date", "Name"]).sum()
                     tran
date       Name          
2012-06-01 jane Doe    72
2012-06-02 jane Doe    75
2013-03-05 john Doe    10
2013-05-05 john Doe   -20