Transpose Pandas DataFrame by select Datetime values in rows

Question

I have a simple Pandas DataFrame that looks like this:

Timestamp              Value
2017-01-03 00:00:00    13.05
2017-01-03 00:01:00    13.70
2017-01-03 00:02:00    13.07
...
2017-01-27 09:57:00    13.46
2017-01-27 09:58:00    13.43
2017-01-27 09:59:00    13.41

How can I transpose it so as to have a single row per day, and columns for specific hours and minutes (say from 0 to 10 hours) (there are no seconds in the Timestamp, they are all zeros).

So far, I can successfully use df[df.index.time < datetime.time(10,0)] to select the rows for each day below hour 10. But if I subsequently do .T to transpose that selection, then the entire DataFrame gets transposed into one very long row. How can I perform that .T operation on batches of rows for each distinct day from the Timestamp? (and also possibly put each Day value in the first column of the transposed dataset?

I would rather avoid custom code and looping, as the dataset is large, so something built-in is much preferred!

Answer 1

Set the index to be a multi-index of date and then time followed by an unstack

df.set_index(
    [df.Timestamp.dt.date, df.Timestamp.dt.time]
).Value.rename_axis([None] * 2).unstack()

Answer 2

You can use pivot_table method:

In [214]: df.assign(d=df.Timestamp.dt.date, t=df.Timestamp.dt.time) \
            .pivot_table(index='d', columns='t', values='Value', fill_value=0) \
            .rename_axis(None, 1)
Out[214]:
            00:00:00  00:01:00  00:02:00  09:57:00  09:58:00  09:59:00
2017-01-03     13.05      13.7     13.07      0.00      0.00      0.00
2017-01-27      0.00       0.0      0.00     13.46     13.43     13.41

PS but I would prefer @piRSquared's solution as it's more elegant...