CODEWITHSUNDEEP
pythonpandasdatetimedataframe
Fanylion
Fanylion

Reputation: 412

Converting pandas date column into seconds elapsed

I have a pandas dataframe of multiple columns with a column of datetime64[ns] data. Time is in HH:MM:SS format. How can I convert this column of dates into a column of seconds elapsed? Like if the time said 10:00:00 in seconds that would be 36000. The seconds should be in a float64 type format.

Example data column

Example data column

Upvotes: 4

Views: 15839

Answers (2)

piRSquared
piRSquared

Reputation: 294228

New Answer
Convert your text to Timedelta

df['Origin Time(Local)'] = pd.to_timedelta(df['Origin Time(Local)'])
df['Seconds'] = df['Origin Time(Local)'].dt.total_seconds()

Old Answer

Consider the dataframe df

df = pd.DataFrame(dict(Date=pd.date_range('2017-03-01', '2017-03-02', freq='2H')))

                  Date
0  2017-03-01 00:00:00
1  2017-03-01 02:00:00
2  2017-03-01 04:00:00
3  2017-03-01 06:00:00
4  2017-03-01 08:00:00
5  2017-03-01 10:00:00
6  2017-03-01 12:00:00
7  2017-03-01 14:00:00
8  2017-03-01 16:00:00
9  2017-03-01 18:00:00
10 2017-03-01 20:00:00
11 2017-03-01 22:00:00
12 2017-03-02 00:00:00

Subtract the most recent day from the timestamps and use total_seconds. total_seconds is an attribute of a Timedelta. We get a series of Timedeltas by taking the difference between two series of Timestamps.

(df.Date - df.Date.dt.floor('D')).dt.total_seconds()
# equivalent to
# (df.Date - pd.to_datetime(df.Date.dt.date)).dt.total_seconds()

0         0.0
1      7200.0
2     14400.0
3     21600.0
4     28800.0
5     36000.0
6     43200.0
7     50400.0
8     57600.0
9     64800.0
10    72000.0
11    79200.0
12        0.0
Name: Date, dtype: float64

Put it in a new column

df.assign(seconds=(df.Date - df.Date.dt.floor('D')).dt.total_seconds())

                  Date  seconds
0  2017-03-01 00:00:00      0.0
1  2017-03-01 02:00:00   7200.0
2  2017-03-01 04:00:00  14400.0
3  2017-03-01 06:00:00  21600.0
4  2017-03-01 08:00:00  28800.0
5  2017-03-01 10:00:00  36000.0
6  2017-03-01 12:00:00  43200.0
7  2017-03-01 14:00:00  50400.0
8  2017-03-01 16:00:00  57600.0
9  2017-03-01 18:00:00  64800.0
10 2017-03-01 20:00:00  72000.0
11 2017-03-01 22:00:00  79200.0
12 2017-03-02 00:00:00      0.0

Upvotes: 6

FdMon
FdMon

Reputation: 121

it would work:

df['time'].dt.total_seconds()

regards

Upvotes: -1

Related Questions