Question

我有以下数据框：


                            id_x  id_y
department         date               
0                  09/2017     1   NaN
1                  01/2018   149   NaN
                   01/2019   112   4.0
                   02/2018   103   1.0
                   02/2019    78   NaN
...                          ...   ...
799                09/2017    57   2.0
                   10/2017    64   3.0
                   11/2017    80   NaN
                   12/2017    79   2.0

这是从数据库数据构建的数据帧的结果，其中按部门和日期对一系列计数进行了运行和分组。

我需要按部门和日期来汇总此数据，但是，我想让日期排在最前面，然后是id的计数。

我想要的输出类似于：

                              9/2017      10/2017
                            id_x   id_y  id_x   id_y
department 
0                              1   NaN    NaN   NaN
1                            NaN   NaN    NaN   NaN
...                          ...   ...    ...   ...
799                           57   2.0     64   3.0

我尝试删除索引，重新索引，熔化数据框并旋转数据框。我可以按“ id_x”和“ id_y”后跟日期的顺序来排序数据框，但这并不是一个很好的解决方案，因为它可能会重复每个id重复36个日期。

我一直在参考以下文档： https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.pivot.html

并测试了以下解决方案的变体（以及其他）：

new_df.melt(new_df, col_level=0, id_vars=['department'], value_vars=['id_x','id_y'])
new_df.reset_index().pivot_table(index="department", columns="date") #I've also tried "date" as values and in brackets outside the parenthesis

Answer 1

重新创建了数据，但是我认为这可以满足您的需求？如果date字段实际上是df中的datetime，则排序将以升序显示数据框。

df=pd.DataFrame({'department':[0,1,1,1,1,799,799,799,799],'date':['09/2017','01/2018','01/2019','02/2018','02/2019','09/2017','10/2017','11/2017','12/2017'],'id_x':[1,149,112,103,78,57,64,80,79],'id_y':[np.NaN,np.NaN,4.0,1.0,np.NaN,2.0,3.0,np.NaN,2.0]})


df=df.set_index('department')
df2=df.pivot(columns='date',values=['id_x','id_y'])        

df3=df2.swaplevel(axis=1)
df3.sort_index(axis=1, level=0, inplace=True)

输出：

date       01/2018      01/2019      02/2018  ... 10/2017 11/2017      12/2017     
              id_x id_y    id_x id_y    id_x  ...    id_y    id_x id_y    id_x id_y
department                                    ...                                  
0              NaN  NaN     NaN  NaN     NaN  ...     NaN     NaN  NaN     NaN  NaN
1            149.0  NaN   112.0  4.0   103.0  ...     NaN     NaN  NaN     NaN  NaN
799            NaN  NaN     NaN  NaN     NaN  ...     3.0    80.0  NaN    79.0  2.0

按日期和组透视计数的熊猫数据框

1 个答案: