Subtract two numpy arrays containing datetime.time

user_7318488 0
Viewed 1844

In my script I get two huge np arrays that I want to subtract. The first numpy array start_per_day with the times the working day started and a second numpy array end_per_day. However I get the below error. Couldn't figure it out, appreciate any help.

Edit: below is a minimal reproducible example where the arrays are artificially were created by me

Edit2: Since I was asked multiple Times about the origina of my two datetime.time numpy arrays, I extract them from pandas groupby with min and max see code below

print(one_employe)
start_per_day = one_employe.groupby('Date')['Time'].min()
end_per_day = one_employe.groupby('Date')['Time'].max()
print(end_per_day.values)
print(start_per_day.values)

The output of the 3 print commands

   ID            Datetime        Date      Time
0  13 2020-03-02 09:43:00  2020-03-02  09:43:00
3  13 2020-03-02 18:22:00  2020-03-02  18:22:00
5  13 2020-03-03 09:51:00  2020-03-03  09:51:00
8  13 2020-03-03 18:28:00  2020-03-03  18:28:00
[datetime.time(18, 22) datetime.time(18, 28)]
[datetime.time(9, 43) datetime.time(9, 51)]

Original Post - Code:

import datetime
import numpy as np
start_per_day = np.array([datetime.time(9, 43),datetime.time(9, 51)])
end_per_day = np.array([datetime.time(18,22),datetime.time(18,29)])

#Attemps that didnt work:
np.datetime64(end_per_day) - np.datetime64(start_per_day)
datetime.timedelta( end_per_day, start_per_day)
end_per_day - start_per_day

Error:

TypeError: unsupported operand type(s) for -: 'datetime.time' and 'datetime.time'

numpy datetime python
1 Answers

You may use datetime.timedelta instead of datetime.time: 

import datetime
import numpy as np

start_per_day = np.array([datetime.timedelta(hours=9, minutes=43), datetime.timedelta(hours=9, minutes=51)])
end_per_day = np.array([datetime.timedelta(hours=18, minutes=22), datetime.timedelta(hours=18, minutes=29)])

delta_per_day = end_per_day - start_per_day

print(delta_per_day[0])
print(delta_per_day[1])

I was looking for a way for subtracting datetime.time objects but I couldn't find any.

Conversion testing (without a loop): 

t1 = np.array([datetime.time(9, 43),datetime.time(9, 51)])
t2 = np.array([datetime.time(18, 22), datetime.time(18, 29)])

start_per_day = np.array([datetime.timedelta(hours=t1[0].hour, minutes=t1[0].minute), datetime.timedelta(hours=t1[1].hour, minutes=t1[1].minute)])
end_per_day = np.array([datetime.timedelta(hours=t2[0].hour, minutes=t2[0].minute), datetime.timedelta(hours=t2[1].hour, minutes=t2[1].minute)])

delta_per_day = end_per_day - start_per_day

print(delta_per_day[0])
print(delta_per_day[1])

Converting array of datetime.time objects to array of datetime.timedelta objects: 

def time_to_delat(t):
    """Convert datetime.time object with hour and minute to datetime.timedelta object"""
    dt = datetime.timedelta(hours=t.hour, minutes=t.minute)
    return dt

start_per_day = np.array([datetime.time(9, 43),datetime.time(9, 51)])
end_per_day = np.array([datetime.time(18, 22), datetime.time(18, 29)])

# Convert array of datetime.time objects to array of datetime.timedelta objects
start_per_day_dt = np.array([time_to_delat(x) for x in start_per_day])
end_per_day_dt = np.array([time_to_delat(x) for x in end_per_day])

delta_per_day = end_per_day_dt - start_per_day_dt

print(delta_per_day[0])
print(delta_per_day[1])
user_49267570
Related