Pandas and groupby count the number of matches in two different columns

Question

I would like to count the number of matches after a groupby in a pandas dataframe.

claim   event   material1   material2
A       X       M1          M2
A       X       M2          M3
A       X       M3          M0
A       X       M4          M4
A       Y       M5          M5
A       Y       M6          M0
B       Z       M7          M0
B       Z       M8          M0

First, I group by the pair claim event and for each of these groups I want to count the number of matches between the columns material1 and material 2

For the group by, I have grouped = df.groupby(['claim', 'event']) but then I don't know how to compare the two new columns.

It should return the following dataframe :

claim   event   matches
A       X       3          
A       Y       1          
B       Z       0          

Do you have any idea how to do that ?

Answer 1

Use isin for compare columns and groupby by columns with aggregate sum, last cast to int and reset_index for columns from MultiIndex:

a = (df['material1'].isin(df['material2']))
df = a.groupby([df['claim'], df['event']]).sum().astype(int).reset_index(name='matches')

Solution with assign to new column:

df['matches'] = df['material1'].isin(df['material2']).astype(int)
df = df.groupby(['claim', 'event'])['matches'].sum().reset_index()

Solutions by @Wen, thank you:

df['matches'] = df['material1'].isin(df['material2']).astype(int)
df = df.groupby(['claim', 'event'], as_index=False)['matches'].sum()

I think it should be slowier in larger DataFrames:

df = (df.groupby(['claim', 'event'])
                  .apply(lambda x : x['material1'].isin(x['material2']).astype(int).sum())
                  .reset_index(name='matches'))

---

print (df)
  claim event  matches
0     A     X        3
1     A     Y        1
2     B     Z        0