Question

我有一个与此类似的数据框：

questions = ['What color?', 'What day?', 'How cold?', 'What color?', 'What color?']
answers = ['red', 'tuesday', '45', 'blue', 'red']
ids = [0, 1, 2, 3, 0]
df = pd.DataFrame({'id': [0, 1, 2, 0, 0], 'questions': questions, 'answers': answers})

 >>> id questions    answers
      0  What color? red
      1  What day?   tuesday
      2  How cold?   45
      0 What color?  blue
      0 What color?  red

我想要这个：

     How cold? What color?  What day?
id          
0       None    red          None
2       None    None         tuesday
3       45      None         None
4       None    blue         None
0       None    red          None

我试过了：

df.pivot(values='answers', index='id', columns='questions')

但是，由于索引中的重复项，pivot始终会导致错误。

Answer 1

您可以使用pivot方法实现这一目标：

df.pivot(columns="questions",values="answers")

输出

    How cold?   What color? What day?
0   NaN          red         NaN
1   NaN          NaN     tuesday
2   45           NaN         NaN
3   NaN         blue         NaN
4   NaN         red          NaN

编辑如果你想保留你拥有的索引，你可以这样做：

new_df = df.pivot(columns="questions",values="answers")
new_df.index = df.index

Answer 2

如果需要重复：

df['g'] = df.groupby('id').cumcount()
df = df.set_index(['id','g', 'questions']).unstack().reset_index(level=1, drop=True)
print (df)
questions How cold? What color? What day?
id                                       
0              None         red      None
0              None        blue      None
0              None         red      None
1              None        None   tuesday
2                45        None      None

用重复索引重构大熊猫数据框架

2 个答案: