CODEWITHSUNDEEP
pythonpandasdataframe
Caglar
Caglar

Reputation: 53

Making a group in dataframe in pandas

I have a list such as

groups = [['Group1', 'A', 'B'], ['Group2', 'C', 'D']]

and a dataframe such as

A 100
B 200
C 300
D 400

I want to make a group sum from the list above to become:

Group 1 300
Group 2 700

How can I do this using python pandas? Needless to say I am a newbie in pandas. Thanks.

Upvotes: 5

Views: 1942

Answers (3)

jezrael
jezrael

Reputation: 862511

You need create dict by lists and then groupby and aggregating sum:

df = pd.DataFrame({'a': ['A', 'B', 'C', 'D'], 'b': [100, 200, 300, 400]})
print (df)
   a    b
0  A  100
1  B  200
2  C  300
3  D  400

groups = [['Group1', 'A', 'B'], ['Group2', 'C', 'D']]

#http://stackoverflow.com/q/43227103/2901002
d = {k:row[0] for row in groups for k in row[1:]}
print (d)
{'B': 'Group1', 'C': 'Group2', 'D': 'Group2', 'A': 'Group1'}

print (df.set_index('a').groupby(d).sum())
          b
Group1  300
Group2  700

Is possible a bit modify solution - if where only column b is aggregate by sum. Last reset_index for convert index to column.

df1 = df.set_index('a').groupby(pd.Series(d, name='a'))['b'].sum().reset_index()
print (df1)
        a    b
0  Group1  300
1  Group2  700

df2 = df.groupby(df['a'].map(d))['b'].sum().reset_index()
print (df2)
        a    b
0  Group1  300
1  Group2  700

Upvotes: 4

piRSquared
piRSquared

Reputation: 294218

Use python 3 unpacking and comprehension to create dictionary. Use that dictionary in a map on first column. Use that mapping to group by.

Consider the list of lists groups and dataframe df

groups = [['Group1', 'A', 'B'], ['Group2', 'C', 'D']]
df = pd.DataFrame(dict(a=list('ABCD'), b=range(100, 401, 100)))

Then:

df.groupby(df.a.map({k: g for g, *c in groups for k in c})).sum()

          b
a          
Group1  300
Group2  700

Upvotes: 1

su79eu7k
su79eu7k

Reputation: 7306

Another option...but seems @jezrael's way is better!

import pandas as pd

groups = [['Group1', 'A', 'B'], ['Group2', 'C', 'D']]

df0 = pd.melt(pd.DataFrame(groups).set_index(0).T)
df1 = pd.read_clipboard(header=None)  # Your example data

df = df1.merge(df0, left_on=0, right_on='value')[['0_y', 1]]
df.columns = ['Group', 'Value']

print df.groupby('Group').sum()


        Value
Group        
Group1    300
Group2    700

Upvotes: 1

Related Questions