Python DataFrame

可以参考下这个， 对ann date分组后 取 report date 最大的记录
D=pd.DataFrame([(1,'a','a'),(2,'a','b'),(3,'b','a'),(4,'b','b')],columns=['F','Ann','Report'])

G=D.groupby('Ann')

G.apply(lambda x:x.sort_values(by='Report',ascending=False).iloc[0])
Out[29]: 
     F Ann Report
Ann              
a    2   a      b
b    4   b      b





import pandas as pd

D=pd.DataFrame([(1,'a','a'),(2,'a','b'),(3,'b','a'),(4,'b','b')],columns=['F','Ann','Report'])

D.set_index('Ann',inplace=True)

D
Out[10]: 
     F Report
Ann          
a    1      a
a    2      b
b    3      a
b    4      b

G=D.groupby(D.index)

G.apply(lambda x:x.sort_values(by='Report',ascending=False).iloc[0])
Out[12]: 
     F Report
Ann          
a    2      b
b    4      b




A=pd.DataFrame([('2000-01-01',10,'1999-01-01')],columns=['ANN','Factor_A','Report_A'])

A.set_index('ANN',inplace=True)

B=pd.DataFrame([('2001-01-01',10,'1993-01-01')],columns=['ANN','Factor_B','Report_B'])

B.set_index('ANN',inplace=True)

A
Out[20]: 
            Factor_A    Report_A
ANN                             
2000-01-01        10  1999-01-01

B
Out[21]: 
            Factor_B    Report_B
ANN                             
2001-01-01        10  1993-01-01

A.join(B,how='outer').sort_index()
Out[23]: 
            Factor_A    Report_A  Factor_B    Report_B
ANN                                                   
2000-01-01      10.0  1999-01-01       NaN         NaN
2001-01-01       NaN         NaN      10.0  1993-01-01