Python Pandas - Convert the Timedelta to a NumPy timedelta64

To convert a Pandas Timedelta to a NumPy timedelta64, use the to_timedelta64() method. This method returns the underlying NumPy representation of the timedelta object, which stores the duration in nanoseconds.

Syntax

timedelta.to_timedelta64()

Creating a Timedelta Object

First, let's create a Timedelta object using a string representation ?

import pandas as pd

# Create a Timedelta object
timedelta = pd.Timedelta('2 days 11 hours 22 min 25 s 50 ms 45 ns')

print("Original Timedelta:")
print(timedelta)
print("Type:", type(timedelta))

Original Timedelta:
2 days 11:22:25.050000045
Type: <class 'pandas._libs.tslibs.timedeltas.Timedelta'>

Converting to NumPy timedelta64

Now let's convert the Timedelta to a NumPy timedelta64 object ?

import pandas as pd

# Create a Timedelta object
timedelta = pd.Timedelta('2 days 11 hours 22 min 25 s 50 ms 45 ns')

# Convert to NumPy timedelta64
numpy_timedelta = timedelta.to_timedelta64()

print("Original Timedelta:")
print(timedelta)
print("\nNumPy timedelta64:")
print(numpy_timedelta)
print("Type:", type(numpy_timedelta))

Original Timedelta:
2 days 11:22:25.050000045

NumPy timedelta64:
213745050000045 nanoseconds
Type: <class 'numpy.timedelta64'>

Working with Different Time Units

You can also create Timedelta objects with different time units and see how they convert ?

import pandas as pd

# Different Timedelta examples
examples = [
    pd.Timedelta('1 day'),
    pd.Timedelta('5 hours'),
    pd.Timedelta('30 minutes'),
    pd.Timedelta('45 seconds')
]

for td in examples:
    numpy_td = td.to_timedelta64()
    print(f"Pandas: {td} ? NumPy: {numpy_td}")

Pandas: 1 days 00:00:00 ? NumPy: 86400000000000 nanoseconds
Pandas: 0 days 05:00:00 ? NumPy: 18000000000000 nanoseconds
Pandas: 0 days 00:30:00 ? NumPy: 1800000000000 nanoseconds
Pandas: 0 days 00:00:45 ? NumPy: 45000000000 nanoseconds

Key Points

• The to_timedelta64() method converts Pandas Timedelta to NumPy timedelta64
• NumPy timedelta64 stores duration in nanoseconds by default
• The conversion preserves the exact duration with nanosecond precision
• This is useful for interoperability between Pandas and NumPy operations

Conclusion

The to_timedelta64() method provides seamless conversion between Pandas Timedelta and NumPy timedelta64 objects. This conversion is essential when working with time-based calculations that require NumPy's underlying representation.