Convert timedelta64 column to Seconds in Pandas DataFrame

# Table of Contents

# Convert timedelta64 column to Seconds in Pandas DataFrame

To convert a timedelta64[ns] column to seconds in a Pandas DataFrame:

Use square brackets to access the specific column.
Use the dt.total_seconds() method to return the total duration of each value in seconds.

main.py

Copied!
import pandas as pd
import numpy as np

df = pd.DataFrame({
    'name': ['Alice', 'Bobby', 'Carl', 'Dan'],
    'job': ['dev', 'web dev', 'accountant', 'dev'],
    'task_duration': [
        np.timedelta64(1, "D"),
        np.timedelta64(30, "m"),
        np.timedelta64(90, "s"),
        np.timedelta64(1, "h")
    ],
})

print(df)

print('-' * 50)

print(df['task_duration'].dt.total_seconds())

The code for this article is available on GitHub

Running the code sample produces the following output.

shell

Copied!
    name         job   task_duration
0  Alice         dev 1 days 00:00:00
1  Bobby     web dev 0 days 00:30:00
2   Carl  accountant 0 days 00:01:30
3    Dan         dev 0 days 01:00:00
--------------------------------------------------
0    86400.0
1     1800.0
2       90.0
3     3600.0
Name: task_duration, dtype: float64

convert timedelta64 ns column to seconds in dataframe

The DataFrame has a task_duration column of type timedelta64[ns].

main.py

Copied!
import pandas as pd
import numpy as np

df = pd.DataFrame({
    'name': ['Alice', 'Bobby', 'Carl', 'Dan'],
    'job': ['dev', 'web dev', 'accountant', 'dev'],
    'task_duration': [
        np.timedelta64(1, "D"),
        np.timedelta64(30, "m"),
        np.timedelta64(90, "s"),
        np.timedelta64(1, "h")
    ],
})

#     name         job   task_duration
# 0  Alice         dev 1 days 00:00:00
# 1  Bobby     web dev 0 days 00:30:00
# 2   Carl  accountant 0 days 00:01:30
# 3    Dan         dev 0 days 01:00:00
print(df)

print(df['task_duration'].dtype)  # 👉️ timedelta64[ns]

The code for this article is available on GitHub

We used the dt.total_seconds() method to convert the column to seconds.

main.py

Copied!
# 0    86400.0
# 1     1800.0
# 2       90.0
# 3     3600.0
# Name: task_duration, dtype: float64
print(df['task_duration'].dt.total_seconds())

The method returns the total duration of each value in seconds.

Make sure to access the .dt namespace on the column before calling total_seconds().

If you need to convert the column in seconds and update the DataFrame, use bracket notation.

main.py

Copied!
import pandas as pd
import numpy as np

df = pd.DataFrame({
    'name': ['Alice', 'Bobby', 'Carl', 'Dan'],
    'job': ['dev', 'web dev', 'accountant', 'dev'],
    'task_duration': [
        np.timedelta64(1, "D"),
        np.timedelta64(30, "m"),
        np.timedelta64(90, "s"),
        np.timedelta64(1, "h")
    ],
})

print(df)

print('-' * 50)

df['task_duration'] = df['task_duration'].dt.total_seconds()

print(df)

The code for this article is available on GitHub

Running the code sample produces the following output.

shell

Copied!
    name         job   task_duration
0  Alice         dev 1 days 00:00:00
1  Bobby     web dev 0 days 00:30:00
2   Carl  accountant 0 days 00:01:30
3    Dan         dev 0 days 01:00:00
timedelta64[ns]
--------------------------------------------------
    name         job  task_duration
0  Alice         dev        86400.0
1  Bobby     web dev         1800.0
2   Carl  accountant           90.0
3    Dan         dev         3600.0

convert timedelta64 column to seconds and update dataframe

The code sample updates the task_duration column from timedelta64[ns] to seconds in place.

You can also add a new column that stores the duration in seconds.

main.py

Copied!
import pandas as pd
import numpy as np

df = pd.DataFrame({
    'name': ['Alice', 'Bobby', 'Carl', 'Dan'],
    'job': ['dev', 'web dev', 'accountant', 'dev'],
    'task_duration': [
        np.timedelta64(1, "D"),
        np.timedelta64(30, "m"),
        np.timedelta64(90, "s"),
        np.timedelta64(1, "h")
    ],
})

print(df)

print('-' * 50)

df['duration_sec'] = df['task_duration'].dt.total_seconds()

print(df)

The code for this article is available on GitHub

Running the code sample produces the following output.

shell

Copied!
    name         job   task_duration
0  Alice         dev 1 days 00:00:00
1  Bobby     web dev 0 days 00:30:00
2   Carl  accountant 0 days 00:01:30
3    Dan         dev 0 days 01:00:00
--------------------------------------------------
    name         job   task_duration  duration_sec
0  Alice         dev 1 days 00:00:00       86400.0
1  Bobby     web dev 0 days 00:30:00        1800.0
2   Carl  accountant 0 days 00:01:30          90.0
3    Dan         dev 0 days 01:00:00        3600.0

convert timedelta64 ns column to seconds add new column

We converted the timedelta64[ns] column to seconds and stored the result in the duration_sec column.

# Convert `timedelta64[ns]` column to seconds by dividing

You can also convert a timedelta64[ns] column to seconds by dividing.

main.py

Copied!
import pandas as pd
import numpy as np

df = pd.DataFrame({
    'name': ['Alice', 'Bobby', 'Carl', 'Dan'],
    'job': ['dev', 'web dev', 'accountant', 'dev'],
    'task_duration': [
        np.timedelta64(1, "D"),
        np.timedelta64(30, "m"),
        np.timedelta64(90, "s"),
        np.timedelta64(1, "h")
    ],
})

print(df)

print('-' * 50)

print(df['task_duration'] / np.timedelta64(1, 's'))

The code for this article is available on GitHub

Running the code sample produces the following output.

shell

Copied!
    name         job   task_duration
0  Alice         dev 1 days 00:00:00
1  Bobby     web dev 0 days 00:30:00
2   Carl  accountant 0 days 00:01:30
3    Dan         dev 0 days 01:00:00
--------------------------------------------------
0    86400.0
1     1800.0
2       90.0
3     3600.0
Name: task_duration, dtype: float64

convert timedelta64 column to seconds by dividing

We simply divided the values in the task_duration column by a timedelta64 object that represents 1 second.

This works in the same way, however, is a bit more implicit than using dt.total_seconds().

If you're working with a Pandas Series object, you can use the following code sample.

main.py

Copied!
series.apply(lambda x: x / timedelta64(1, 's'))

The Series.apply() method invokes a function on the values of a Series object.

If you don't want to use the Series.apply() method, use numpy.divide().

main.py

Copied!
np.divide(series, np.timedelta64(1, 's'))

# Additional Resources

You can learn more about the related topics by checking out the following tutorials:

I wrote a book in which I share everything I know about how to become a better, more efficient programmer.

You can use the search field on my Home Page to filter through all of my articles.

Convert timedelta64 column to Seconds in Pandas DataFrame

# Table of Contents