How to bin a time only timestamp in pandas?

Question

I have a dataframe df

user    timestamp  amount
us67    15:59:07   87
us90    17:12:19   10
us12    03:23:16   17

print(df.timestamp[0])
>>> datetime.time(15,59,7)

I want to bin all the times into 1 hour intervals, so 24 intervals in total. However, I get a TypeError

df['timestamp'] = pd.cut(x=df['timestamp'], bins=24)
>>> TypeError: unsupported operand type(s) for +: 'datetime.time' and 'float'

The method does work, however, if the date is included in the timestamp column, but I want to ignore the date and only keep the times (for plotting later on):

user    timestamp                 amount
us67    2018-04-29 15:59:07.455   87
us90    2018-04-29 17:12:19.128   10
us12    2018-04-29 03:23:16.890   17

print(df.timestamo[0])
>>> Timestamp('2018-04-29 15:59:07.455000')

df['timestamp'] = pd.cut(x=df['timestamp'], bins=24)

With the format above for timestamp, the binning works. However I don't want the year and date in the timestamp or in the intervals. I want to focus only on the time of day.

Is there a way I can bin timestamp using only time of day? Ultimately the goal here is to plot a timeseries of df (timestamp vs. amount) using the hour of day only and not the date - so if there is a better way to do that, please suggest it.

Answer 1

I would create a column with my binned hours using dt.hour

So

df["binned_hours"] = pd.cut(df.timestamp.dt.hour, bins=24)