Pandas-基于参考字典重复数据框列答案

【问题标题】：Pandas- repeat dataframe columns based on a reference dictPandas-基于参考字典重复数据框列
【发布时间】：2017-06-22 21:00:26
【问题描述】：

我需要根据参考字典重命名和重复我的数据框列。下面我创建了一个虚拟数据框：

rawdata= {'id':['json','molly','tina','jake','molly'],'entity':['present','absent','absent','present','present'],'entity2':['present','present','present','absent','absent'],'entity3':['absent','absent','absent','present','absent']}
df= pd.DataFrame(rawdata)
df.set_index('id')

        entity  entity2  entity3
id                              
json   present  present   absent
molly   absent  present   absent
tina    absent  present   absent
jake   present   absent  present
molly  present   absent   absent

现在我有以下示例字典：

ref_dict= {'entity':['entity_exp1'],'entity2':['entity2_exp1','entity2_exp2'],'entity3':['entity3_exp1','entity3_exp2','entity3_exp3']}

我现在需要根据 dict 值替换列名，如果一列有多个值，则应该重复该列。以下是我想要的数据框：

       entity_exp1  entity2_exp1 entity2_exp2 entity3_exp1 entity3_exp2 entity3_exp3
id                      
json    present      present      present      absent      absent    absent
molly   absent       present      present      absent      absent    absent
tina    absent       present      present      absent      absent    absent
jake    present      absent       absent       present     present   present
molly   present      absent       absent       absent      absent    absent

【问题讨论】：

感谢您排除我的回答。也可以随意投票。
感谢 piRSquared。你总是有最令人惊叹的解决方案。

标签： python pandas dataframe

【解决方案1】：

选项 1
在字典理解上使用pd.concat

pd.concat({k: df[v] for v, l in ref_dict.items() for k in l}, axis=1)

      entity2_exp1 entity2_exp2 entity3_exp1 entity3_exp2 entity3_exp3 entity_exp1
id                                                                                
json       present      present       absent       absent       absent     present
molly      present      present       absent       absent       absent      absent
tina       present      present       absent       absent       absent      absent
jake        absent       absent      present      present      present     present
molly       absent       absent       absent       absent       absent     present

选项 2
切片数据框并重命名列

repeats = df.columns.map(lambda x: len(ref_dict[x]))
d1 = df.reindex_axis(df.columns.repeat(repeats), 1)
d1.columns = df.columns.map(ref_dict.get).values.sum()
d1

      entity_exp1 entity2_exp1 entity2_exp2 entity3_exp1 entity3_exp2 entity3_exp3
id                                                                                
json      present      present      present       absent       absent       absent
molly      absent      present      present       absent       absent       absent
tina       absent      present      present       absent       absent       absent
jake      present       absent       absent      present      present      present
molly     present       absent       absent       absent       absent       absent

【讨论】：

【解决方案2】：

您可以使用 dict 键作为列名重新索引您的 df，然后使用 dict 的值重命名列。

df_new = df.reindex(columns=sum([[k]*len(v) for k,v in ref_dict.items()],[]))
df_new.columns=sum(ref_dict.values(),[])
df_new
Out[573]: 
  entity_exp1 entity2_exp1 entity2_exp2 entity3_exp1 entity3_exp2 entity3_exp3
0     present      present      present       absent       absent       absent
1      absent      present      present       absent       absent       absent
2      absent      present      present       absent       absent       absent
3     present       absent       absent      present      present      present
4     present       absent       absent       absent       absent       absent

【讨论】：

【解决方案3】：

你可以简单地循环：

rawdata= {'id':['json','molly','tina','jake','molly'],
          'entity':['present','absent','absent','present','present'],
          'entity2':['present','present','present','absent','absent'],
          'entity3':['absent','absent','absent','present','absent']}
df= pd.DataFrame(rawdata)
df.set_index('id')
ref_dict= {'entity':['entity_exp1'],
           'entity2':['entity2_exp1','entity2_exp2'],
           'entity3':['entity3_exp1','entity3_exp2','entity3_exp3']}

# here comes the new part:
df2 = pd.DataFrame()
for key, val in sorted(ref_dict.items()):
    for subval in val:
        df2[subval] = df[key]

df2['id'] = df['id']
df2.set_index('id', inplace=True)

print(df2)
      entity_exp1 entity2_exp1 entity2_exp2 entity3_exp1 entity3_exp2  entity3_exp3  
id                                                                      
json      present      present      present       absent       absent        absent   
molly      absent      present      present       absent       absent        absent   
tina       absent      present      present       absent       absent        absent   
jake      present       absent       absent      present      present       present    
molly     present       absent       absent       absent       absent        absent

【讨论】：

【解决方案4】：

对于df 中的每一列，您可以在ref_dict 中查找新列的数量并为它们创建new column，最后删除旧列。您可以尝试以下操作：

# for key, value in ref_dict where old column and new columns are 
for old_column,new_columns in ref_dict.items():
    for new_column in new_columns:  # for each new_column in new_columns defined
        df[new_column] = df[old_column] # the content remains same as old column
    del df[old_column]  # now remove the old column

【讨论】：