将dataframe列转换为嵌套python字典

Question

我想用pandas数据框第2列（源代码）和第3列（说明）以及按第1列分组（标题）创建python字典 另外，我希望只获取提供的标题的值 标题=['test1'，'test2']

   title  source description
1  Test1    ABC  description1
2  Test2    ABC  description2
3  Test2    DEF  description3
4  Test3    XYZ  description4

output = {'Test1':{'ABC':'description1'},'Test2':{'ABC':'description2':'DEF':'description3'}

Answer 1

您可以按dataframe w.r.t.标题分组，然后使用python zip函数创建包含源代码和描述的内部字典。请查找以下相同的代码

final_dict=dict()
all_groups = df.groupby('title')
for title in titles: 
    title_group = all_groups.get_group(title)
    source_desc=dict(zip(title_group.source, title_group.description))
    final_dict[title_group] = source_desc
print(final_dict)

Answer 2

试试这个

result = {}

filter_ = ['Test1','Test2']

for x in df[df['title'].isin(filter_)].to_dict(orient='records'):
    result.setdefault(x['title'], {}).update({x['source']: x['description']})

---

{'Test1': {'ABC': 'description1'}, 'Test2': {'ABC': 'description2', 'DEF': 'description3'}}

Answer 3

首先使用^{}和^{}作为过滤器，然后使用^{}和lambda函数作为DICT的Series和最后一个^{}：

titles = ['Test1','Test2']

d = (df[df['title'].isin(titles)]
       .groupby('title')[['source','description']]
       .apply(lambda x: dict(x.to_numpy()))
       .to_dict())
print (d)
{'Test1': {'ABC': 'description1'}, 'Test2': {'ABC': 'description2', 'DEF': 'description3'}}