python - 如何在不丢失熊猫记录顺序的情况下连接连续行？

Question

当连续行中变量的类别相同时，我试图根据分类变量连接两行。下面是我的数据，例如：

   SNo  user    Text
0   1   Sam     Hello
1   1   John    Hi
2   1   Sam     How are you?
3   1   John    I am good
4   1   John    How about you?
5   1   John    How is it going?
6   1   Sam     Its going good
7   1   Sam     Thanks
8   2   Mary    Howdy?
9   2   Jake    Hey!!
10  2   Jake    What a surprise
11  2   Mary    Good to see you here :)
12  2   Jake    Ha ha. Hectic life
13  2   Mary    I know right..
14  2   Mary    How's Amy doing?
15  2   Mary    How are the kids?
16  2   Jake    All is good! :)

在这里，如果我之前的 user 列值与我当前的 user 列值相同，但与该列中的下一个值不同，那么，我将连接列 Text 中的值那个用户。我需要这样做，直到该特定用户不再多次出现。下面给出了一个示例输出：

SNo user    Text
1   Sam     Hello
1   John    Hi
1   Sam     How are you?
1   John    I am good-How about you?-How is it going?
1   Sam     Its going good-Thanks
2   Mary    Howdy?
2   Jake    Hey!!-What a surprise
2   Mary    Good to see you here :)
2   Jake    Ha ha. Hectic life
2   Mary    I know right..-How's Amy doing?-How are the kids?
2   Jake    All is good! :)

我尝试使用df.groupby() 然后.agg() 完成连接，但无法对其应用上述条件。因此，输出将组合所有出现的用户进行聊天。

df = sample_data.groupby(["SNo","user"]).agg({'Text': '-'.join}).reset_index() # incorrect though
df

此外，我试图避免像瘟疫一样的for 循环并尝试矢量化解决方案。

---

样本数据：

data_dict = {'S. No.': [1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2], 'user': ['Sam', 'John', 'Sam', 'John', 'John', 'John', 'Sam', 'Sam', 'Mary', 'Jake', 'Jake', 'Mary', 'Jake ', 'Mary', 'Mary', 'Mary', 'Jake'], 'Text': ['Hello', 'Hi', 'How are you?', 'I am good', 'How about you?', 'How is it going?', 'Its going good', 'Thanks', 'Howdy?', 'Hey!!', 'What a surprise', 'Good to see you here :)', 'Ha ha. Hectic life', 'I know right..', "How's Amy doing?", 'How are the kids?', 'All is good! :)']}

sample_data = pd.DataFrame(data_dict)

Answer 1

您希望将user 与其shift 和cumsum 进行比较以进行更改。然后你可以分组：

blocks = df['user'].ne(df['user'].shift()).cumsum()
(df.groupby(['SNo', blocks])
  .agg({'user':'first','Text': '-'.join})
  .reset_index('user', drop=True)
)

输出：

     user                                               Text
SNo                                                         
1     Sam                                              Hello
1    John                                                 Hi
1     Sam                                       How are you?
1    John          I am good-How about you?-How is it going?
1     Sam                              Its going good-Thanks
2    Mary                                             Howdy?
2    Jake                              Hey!!-What a surprise
2    Mary                            Good to see you here :)
2    Jake                                 Ha ha. Hectic life
2    Mary  I know right..-How's Amy doing?-How are the kids?
2    Jake                                    All is good! :)