Question

我有一个数据框df为：

Acct_Id  Acct_Nm   Srvc_Id   Phone_Nm   Phone_plan_value   Srvc_Num
51       Roger     789       Pixel      30                 1
51       Roger     800       iPhone     25                 2
51       Roger     945       Galaxy     40                 3
78       Anjay     100       Nokia      50                 1
78       Anjay     120       Oppo       30                 2
32       Rafa      456       HTC        35                 1

我想转换数据帧，以便每个Acct_Id和Acct_Nm可以有1行，如下所示：

    Acct_Id   Acct_Nm    Srvc_Num_1                             Srvc_Num_2                              Srvc_Num_3
                         Srvc_Id   Phone_Nm   Phone_plan_value  Srvc_Id   Phone_Nm   Phone_plan_value   Srvc_Id   Phone_Nm   Phone_plan_value
          51  Roger      789       Pixel      30                800       iPhone     25                 945       Galaxy     40
          78  Anjay      100       Nokia      50                120       Oppo       30
          32  Rafa       456       HTC        35

我不确定如何在熊猫中实现同样的目标。

Answer 1

更像一个pivot问题，但需要swaplevel和sort_index

df.set_index(['Acct_Id','Acct_Nm','Srvc_Num']).\
   unstack().\
   swaplevel(1,0,axis=1).\
   sort_index(level=0,axis=1).add_prefix('Srvc_Num_')


Out[289]: 

Srvc_Num               Srvc_Num_1                                             \
                Srvc_Num_Phone_Nm Srvc_Num_Phone_plan_value Srvc_Num_Srvc_Id   
Acct_Id Acct_Nm                                                                
32      Rafa                  HTC                      35.0            456.0   
51      Roger               Pixel                      30.0            789.0   
78      Anjay               Nokia                      50.0            100.0   
Srvc_Num               Srvc_Num_2                                             \
                Srvc_Num_Phone_Nm Srvc_Num_Phone_plan_value Srvc_Num_Srvc_Id   
Acct_Id Acct_Nm                                                                
32      Rafa                 None                       NaN              NaN   
51      Roger              iPhone                      25.0            800.0   
78      Anjay                Oppo                      30.0            120.0   
Srvc_Num               Srvc_Num_3                                             
                Srvc_Num_Phone_Nm Srvc_Num_Phone_plan_value Srvc_Num_Srvc_Id  
Acct_Id Acct_Nm                                                               
32      Rafa                 None                       NaN              NaN  
51      Roger              Galaxy                      40.0            945.0  
78      Anjay                None                       NaN              NaN

这是pivot_table

pd.pivot_table(df,index=['Acct_Id','Acct_Nm'],columns=['Srvc_Num'],values=['Phone_Nm','Phone_plan_value','Srvc_Id'],aggfunc='first')

Answer 2

怎么样：

df.set_index(['Acct_Id', 'Acct_Nm', 'Srvc_Num']).unstack().swaplevel(0, 1, axis = 1).sort_index(axis = 1)

Pandas数据透视表不进行汇总

2 个答案: