CODEWITHSUNDEEP
pythonpython-2.7pandas
MJP
MJP

Reputation: 1597

How to merge 2 complicated data frames in python pandas?

I have 2 pandas data frames. 

dictionary1 = {'match_up' : ['1985_1116_1234' , '1985_1116_1475', '1985_1234_1172', '1985_1475_2132',  '1985_1242_1325'], \
               'result': [1, 1, 0, 0, 1], 'year':[1985,1985,1985,1985,1985]  }


dictionary2 = {'team' : [1234 , 1475,  2132, 1172, 1242, 1116 , 1325], 'win_A_B': [0.667, 0.636, 0.621, 0.629, 0.615,0.943, 0.763], \
               'year':[1985,1985,1985,1985,1985,1985,1985] }

df1 = pd.DataFrame(dictionary1)

df2 = pd.DataFrame(dictionary2)

df1:
           match_up     result  year
    0   1985_1116_1234    1     1985
    1   1985_1116_1475    1     1985
    2   1985_1234_1172    0     1985
    3   1985_1475_2132    0     1985
    4   1985_1242_1325    1     1985

df2:
    team      win_A_B    year
    1234      0.667      1985
    1475      0.636      1985 
    2132      0.621      1985
    1172      0.629      1985
    1242      0.615      1985
    1116      0.943      1985
    1325      0.763      1985

The column value in data frame df1 is a matchup of the column team in the data frame df2. The column team in df2 are all unique values.

I need to combine the above 2 data frames in the following manner:

           match_up     result  year   team_A   team_B    win_A    win_B
    0   1985_1116_1234    1     1985    1116      1234     0.943    0.667    
    1   1985_1116_1475    1     1985    1116       1475    0.943     0.636
    2   1985_1234_1172    0     1985    1234       1172    0.667     0.629
    3   1985_1475_2132    0     1985    1475       2132    0.636    0.621
    4   1985_1242_1325    1     1985    1242       1325    0.615    0.763

I know I have already asked similar questions in pandas. I am new to pandas, so please bear with me if I ask some questions like this.

Upvotes: 4

Views: 94

Answers (1)

cge
cge

Reputation: 9890

The following will work:

d_teams=pd.DataFrame( [[int(y) for y in x.split('_')[1:]] \
            for x in df1.match_up], columns=('team_A', 'team_B') )
merged=pd.concat((df1,d_teams),axis=1)
df2i=df2.set_index('team')
merged['win_A']=df2i.ix[merged.team_A].reset_index().win_A_B
merged['win_B']=df2i.ix[merged.team_B].reset_index().win_A_B

First, we create d_teams, which is a DataFrame that is composed of the match_up column, split by '_', and turned into ints. We throw away the year, because it's already included in df1, and just keep team_A and team_B. Then we create a merged dataframe by concatenating this with df1.

Next, we create df2i, which is df2 indexed by team. Then we can just index with merged.team_A or merged.team_B to get the win values. However, we don't want the results to be indexed by those teams, so we reset the index first.

Upvotes: 2

Related Questions