从 pandas df 中的下一行获取值并插入到上一行

Question

I have a large q/a dataset, which always contains 3 rows per unique id /they are already grouped.

id	from	to	text
1	max		hi
1	charls	max	hey !
1	alex	max	howdi !
2	sandy		hi
2	sandy		hey??!
2	sam	sandy	jeah?

I want only the text values and compress each three rows into one / drop the other two rows.

id	author_text_1	author_text_2	author_text_3	reply_text_1	reply_text_2
1	hi			hey !	howdi !
2	hi	hey??!		jeah?

I tried for each group of 3 rows, to drop the last two rows and add the values from each 'text' column to the corresponding 'author_text' and 'reply_text' columns.

I am struggling with implementing the case that the author could make the question post and 2 more posts which would fill the three rows.

Answer 1

You need to create the column as the new key for pivot

df['key'] = np.where(df.to.isna(),'author_text_','reply_text_')
df['key'] += df.groupby(['id','key']).cumcount().add(1).astype(str)
out = df.pivot('id','key','text')
Out[254]: 
key author_text_1 author_text_2 reply_text_1 reply_text_2
id                                                       
1          NaN hi           NaN         hey!       howdi!
2              hi        hey!?!        jeah?          NaN