使用数据透视表跟踪用户聊天记录

Question

我有一个跟踪多个客户和代表聊天的平台。我想创建包含单个客户的完整有序聊天历史记录的数据框。

这里是a link to the example data

为了快速参考：

Convo Room      Date        Message Order       User ID     Role        Chat contents
A1      3-Oct-17        1       JOHN        CUSTOMER        Hi, can you help?
A1      4-Oct-17        2       ALICE       REP     Sure, what's up?
A1      5-Oct-17        3       JOHN        CUSTOMER        I have warts.
A1      6-Oct-17        4       JOHN        CUSTOMER        Please don't hang up, it's just warts.
B1      7-Oct-17        1       JOHN        CUSTOMER        Hi, can YOU help?
B1      8-Oct-17        2       MARY        REP     Sure, I heard about Alice.
B1      9-Oct-17        3       MARY        REP     I also have warts.
B1      10-Oct-17       4       JOHN        CUSTOMER        Oh, nevermind then, gotta go.
C1      7-Oct-17        1       JIM     CUSTOMER        Hi, can you help?
C1      8-Oct-17        2       ALICE       REP     Maybe, what's up?
C1      9-Oct-17        3       JIM     CUSTOMER        Not warts.
C1      10-Oct-17       4       ALICE       REP     Good, that's the only thing I cannot handle.
D1      15-Oct-17       1       JOHN        CUSTOMER        Hi, pls help. Warts.
D1      16-Oct-17       2       JUDE        REP     Perfect, I cure them!
D1      17-Oct-17       3       JUDE        REP     …with fire.
D1      18-Oct-17       4       JUDE        REP     Are you still there? Dang, lost another one.

在我看来，第一步是使用数据透视表对数据进行排序。接下来，我可以专注于将聊天分成数据框以进行情绪分析或其他指标。

我相信我已经很接近了，但我总是把排序的一部分弄错。

到目前为止我所拥有的：

df = test.pivot_table(index=['Role', 'User ID', 'Date', 'Convo Room', 'Message Order'],columns=["Role"],aggfunc='first')
df.head()

返回以下内容： 使用 Excel，我相信这通常是我想要的，尽管我确信有很多方法可以将其可视化：

Answer 1

你在正确的轨道上，只需要确保你将正确的列传递给pivot_table

df.pivot_table(index=[ 'User ID', 'Convo Room', 'Message Order'],columns=["Role"],values='Chat contents',aggfunc='first')