Taking Median of two datetime values or columns

问题

For the below data I want to take the middle value or the middle time of the first two timestamps in each row and then subract that third timestamp

What would be the best way to take the median value or middle datime of two timestamps?

the output expected is in minutes the difference of two timestamps.

It is the median or mean of the first two minus the third timestamp.

it is the middle value or timestamp of 2018-12-21 23:31:24.615 and 2018-12-21 23:31:26.659.

Once I have that value I want to subtract the third timestamp of 2018-12-21 23:31:27.975. The output would represent a value of minutes.

回答1:

If you just want the middle value of the datetime column, you can do this:

df['linked_trip_pickup_departed_time'].astype('datetime64[ns]').quantile(.5)
df['pickup_departed_time_utc'].astype('datetime64[ns]').quantile(.5)

This will give you the median for each datetime column. Now, you can subtract it.

回答2:

Assuming the df looks like:

df = pd.DataFrame(data={'time1':['2018-12-21 23:31:24.615','2018-12-22 01:33:26.015'],'time2':['2018-12-21 23:31:26.659','2018-12-22 01:33:32.865'],'time3':['2018-12-21 23:31:27.975','2018-12-22 01:59:05.136']})

    time1                   time2                   time3
0   2018-12-21 23:31:24.615 2018-12-21 23:31:26.659 2018-12-21 23:31:27.975
1   2018-12-22 01:33:26.015 2018-12-22 01:33:32.865 2018-12-22 01:59:05.136

Convert 'to_datetime'

df[['time1','time2','time3']] = df[['time1','time2','time3']].apply(pd.to_datetime,errors='coerce')

creating a column having the average of the first 2 columns:

my_list= []
for i in df.index:
    my_list.append(pd.to_datetime((df['time1'][i].value + df['time2'][i].value)/2.0))
df['avg'] = my_list

or simply :

df['avg'] = [(pd.to_datetime((df['time1'][i].value + df['time2'][i].value)/2.0)) for i in df.index]

finding difference of column3 and avg:

(df.time3-df.avg).astype('timedelta64[m]')

output:

0     0.0
1    25.0
dtype: float64

P.S : you have to replace columns time1,time2 and time3 with the column names in your dataframe.

来源：https://stackoverflow.com/questions/53940340/taking-median-of-two-datetime-values-or-columns

标签

python

pandas

datetime

timestamp

timedelta