Question

我有两个看起来像

的数据框

   team  points
0     1     2.5
1     2     3.2
2     5     5.8
3     3     2.8
4     4     1.9

和：

   team1  team2
0      1      5
1      2      4
2      3      1

预期的输出应该给我一个与获胜者更多的专栏（更多分）：

team1   team2   winner
 1         5       5
 2         4       2
 3         1       3

Answer 1

这是使用applymap，df.idxmax()和df.lookup的方法：

s=df2.applymap(df1.set_index('team')['points'].get).idxmax(1)

或者由@ user3483203提供更好的选择

s=df2.stack().map(df1.set_index('team')['points']).unstack().idxmax(1)
#s.tolist() gives ['team2', 'team1', 'team1']
df2['winner']=df2.lookup(s.index,s)
print(df2)

   team1  team2  winner
0      1      5       5
1      2      4       2
2      3      1       3

Answer 2

尝试避免使用applymap并使用lookup + reshape

x = df.set_index('team').lookup(df2.values.ravel('F'), ["points"]*df2.size)
                        .reshape(df2.shape, order='F')
                        .argmax(1)

df2['winner'] = df2.lookup(df2.index, df2.columns[x])

   team1  team2  winner
0      1      5       5
1      2      4       2
2      3      1       3

Answer 3

仅使用pandas.Series.map，DataFrame.stack和DataFrame.unstack的替代解决方案：

df_match['winner']=df_match.stack().map(df.set_index('team')['points']).unstack().max(axis=1).map(df.set_index('points')['team'])

print(df_match)

   team1  team2  winner
0      1      5       5
1      2      4       2
2      3      1       3

Answer 4

我的“简单”解决方案：

df3= df2.replace(df1.set_index("team").points.to_dict()) 
df2["winner"]= np.where(df3.team1>=df3.team2,df2.team1,df2.team2)

将数据框的两列与新的数据框列进行比较

4 个答案: