Python pandas: group by two columns, sum on one but not on the others

Question

I have a dataframe, df:

Location	Category	Species	Number	x	y
a	Duiker	P monticola	3	9.1	-2.1
a	Duiker	C callipygus	6	9.1	-2.1
b	Duiker	C callipygus	4	9.2	-2.2
b	Carnivore	G servalina	2	9.2	-2.2
b	Carnivore	G maculata	3	9.2	-2.2
c	Carnivore	C aurata	1	9.3	-2.3

I want to convert df into the following format with these conditions:

1. remove the Species column
2. group by Category AND Location
3. in grouping, aggregate Number values
4. but do not aggregate columns x & y (same for any Location)

So that df:

Location	Category	Number	x	y
a	Duiker	9	9.1	-2.1
b	Duiker	4	9.2	-2.2
b	Carnivore	5	9.2	-2.2
c	Carnivore	1	9.3	-2.3

My current method would be to:

1. split df into df1 and df2, where

df1 = df[['Location', 'Category', 'Number']].copy()
df2 = df[['Location', 'x', 'y']].copy()

2. group and sum df1 by Location and Category

df1 = df1.groupby('Location', 'Category']).sum()

3. Merge intersect df1 and df2 on Location and Category

df3 = pd.merge(df1, df2, how = 'inner', on = ['Location'])

Instead on step-3. I get a df3 where the Category column is removed and the Locations are no longer grouped, like this:

Location	Number	x	y
a	9	9.1	-2.1
a	9	9.1	-2.1
b	4	9.2	-2.2
b	4	9.2	-2.2
c	1	9.3	-2.3

I'm a bit lazy and a bit stuck, can someone throw me a bone? and perhaps make my coding more efficient in the process. Thank you in advance.

Answer 1

Specify the columns and functions in agg :

• sum the "Number" column
• keep the first values in columns "x" and "y"

>>> df.groupby(["Location","Category"],as_index=False).agg({"Number":"sum","x":"first","y":"first"})

  Location   Category  Number    x    y
0        a     Duiker       9  9.1 -2.1
1        b  Carnivore       5  9.2 -2.2
2        b     Duiker       4  9.2 -2.2
3        c  Carnivore       1  9.3 -2.3