I'm looking for a way to make a transformation like below, the number of rows is the same for every unique id_, timestamp value is also the same for each unique id_. here is a portion from the original df
timestamp id_ value
0 2020-05-01 a 0
1 2020-05-02 a 1
2 2020-05-03 a 2
3 2020-05-04 a 3
4 2020-05-05 a 4
5 2020-05-06 a 5
6 2020-05-07 a 6
7 2020-05-08 a 7
8 2020-05-09 a 8
9 2020-05-10 a 9
10 2020-05-01 b 10
11 2020-05-02 b 11
12 2020-05-03 b 12
13 2020-05-04 b 13
14 2020-05-05 b 14
15 2020-05-06 b 15
16 2020-05-07 b 16
17 2020-05-08 b 17
18 2020-05-09 b 18
19 2020-05-10 b 19
to
a b
2020-05-01 0 10
2020-05-02 1 11
2020-05-03 2 12
2020-05-04 3 13
2020-05-05 4 14
2020-05-06 5 15
2020-05-07 6 16
2020-05-08 7 17
2020-05-09 8 18
2020-05-10 9 19
My current way is as below
df_gb = df.groupby('id_')
ids = df_gb.size().index.tolist()
tem_dict = dict()
index_candidate = df.timestamp.dt.date.unique().tolist()
for ts_id in ids:
tem_dict[ts_id] = df_gb.get_group(ts_id).value.tolist()
res = pd.DataFrame(tem_dict).set_index(index_candidate)
I hope to use built-in functions in pandas since the performance reason. Thanks for your help.
