如何在熊猫中 dcast 数据框

Question

我在熊猫中有以下数据框

  key       no      lpm
  ab_12     1       12 
  ab_12     2       11
  ab_12     3       11
  ac_12     1       12 
  ac_12     2       11
  ac_12     4       11
  ad_12     1       12 
  ad_12     2       11
  ad_12     3       11

我想要的数据框如下

  key    no_1  no_2   no_3             no_4
  ab_12  12    11     11               does not exist
  ac_12  12    11     does not exist   11
  ad_12  12    11     11               does not exist 

我正在熊猫中做以下工作，但它没有给我我需要的东西。

  df= df.melt('key').groupby(['key', 'value']).unstack(fill_value='Does not exist')

Answer 1

set_index与unstack和 一起使用add_prefix：

df = df.set_index(['key', 'no'])['lpm'].unstack(fill_value='Does not exist').add_prefix('no_')
print (df)
no    no_1 no_2            no_3            no_4
key                                            
ab_12   12   11              11  Does not exist
ac_12   12   11  Does not exist              11
ad_12   12   11              11  Does not exist

如果解决方案因为成对重复而不起作用key，no则必须聚合：

df = (df.groupby(['key', 'no'])['lpm']
        .mean()
        .unstack(fill_value='Does not exist')
        .add_prefix('no_'))

或者：

df = (df.pivot_table(index='key', 
                    columns='no', 
                    values='lpm', 
                    fill_value='Does not exist',
                    aggfunc='mean').add_prefix('no_'))

编辑：对于后缀添加add_suffix：

df = (df.set_index(['key', 'no'])['lpm']
        .unstack(fill_value='Does not exist')
        .add_prefix('no_')
        .add_suffix('_lpm'))
print (df)
no    no_1_lpm no_2_lpm        no_3_lpm        no_4_lpm
key                                                    
ab_12       12       11              11  Does not exist
ac_12       12       11  Does not exist              11
ad_12       12       11              11  Does not exist