將多行合並到數據框列的一行

Question

我當前的數據框如下所示，

              0           1         2
0  HA-567034786  AB-1018724      None
1    AB-6348403  HA-7298656      None

使用apply() ，我就像這樣

def make_dict(row):
    s = set(x for x in row if x)
    return {x: list(s - {x}) for x in s}

result = df.apply(make_dict, axis=1).to_frame(name = 'duplicates')

                                duplicates
1    {'HA-567034786': ['AB-1018724'],'AB-1018724':['HA-567034786']}                                                                            
2    {'AB-6348403': ['HA-7298656'],'HA-7298656':['AB-6348403']}   

現在，我堅持要使其成為一個單一的三維字典，如下所示，

{
  'HA-567034786': ['AB-1018724'],'AB-1018724':['HA-567034786'],
  'AB-6348403': ['HA-7298656'],'HA-7298656':['AB-6348403']
}  

Answer 1

相反， apply使用字典理解與扁平化：

print (df)
              0           1
0  HA-567034786  AB-1018724
1    AB-6348403  HA-7298656

def make_dict(row):
    s = set(x for x in row if x)
    return {x: list(s - {x}) for x in s}

result = {k:v for x in df.values for k, v in make_dict(x).items()}

print (result)
{'HA-567034786': ['AB-1018724'],
 'AB-1018724': ['HA-567034786'], 
 'HA-7298656': ['AB-6348403'],
 'AB-6348403': ['HA-7298656']}

另一種解決方案與apply ：

result = {k:v for x in df.apply(make_dict, axis=1) for k, v in x.items()}

Answer 2

您也可以使用collections.ChainMap（）將所有字典歸為一組，如下所示：

from collections import ChainMap
res =dict(ChainMap(*result['duplicates']))