Python Pandas - Return numpy array of python datetime.date objects

To return a numpy array of Python datetime.date objects from a Pandas DatetimeIndex, use the date property. This property extracts only the date part from timestamps, removing timezone information and returning standard Python date objects.

Syntax

datetimeindex.date

Where datetimeindex is a Pandas DatetimeIndex object.

Creating a DatetimeIndex

First, let's create a DatetimeIndex with timezone information ?

import pandas as pd

# Create DatetimeIndex with 3 periods and nanosecond frequency
datetimeindex = pd.date_range('2021-10-20 02:30:50', periods=3, tz='Australia/Sydney', freq='ns')

print("DateTimeIndex...")
print(datetimeindex)

DateTimeIndex...
DatetimeIndex(['2021-10-20 02:30:50+11:00',
               '2021-10-20 02:30:50.000000001+11:00',
               '2021-10-20 02:30:50.000000002+11:00'],
              dtype='datetime64[ns, Australia/Sydney]', freq='N')

Extracting Date Objects

Now extract the date part using the date property ?

import pandas as pd

datetimeindex = pd.date_range('2021-10-20 02:30:50', periods=3, tz='Australia/Sydney', freq='ns')

# Extract numpy array of datetime.date objects
date_array = datetimeindex.date

print("The numpy array (date part)...")
print(date_array)
print(f"\nType of array: {type(date_array)}")
print(f"Type of first element: {type(date_array[0])}")

The numpy array (date part)...
[datetime.date(2021, 10, 20) datetime.date(2021, 10, 20)
 datetime.date(2021, 10, 20)]

Type of array: <class 'numpy.ndarray'>
Type of first element: <class 'datetime.date'>

Complete Example

Here's a comprehensive example showing the conversion process ?

import pandas as pd

# DatetimeIndex with period 3 and frequency as nanoseconds
# The timezone is Australia/Sydney
datetimeindex = pd.date_range('2021-10-20 02:30:50', periods=3, tz='Australia/Sydney', freq='ns')

# Display DateTimeIndex
print("DateTimeIndex...\n", datetimeindex)

# Returns only the date part of Timestamps without timezone information
print("\nThe numpy array (date part)..\n", datetimeindex.date)

DateTimeIndex...
 DatetimeIndex(['2021-10-20 02:30:50+11:00',
               '2021-10-20 02:30:50.000000001+11:00',
               '2021-10-20 02:30:50.000000002+11:00'],
              dtype='datetime64[ns, Australia/Sydney]', freq='N')

The numpy array (date part)..
 [datetime.date(2021, 10, 20) datetime.date(2021, 10, 20)
 datetime.date(2021, 10, 20)]

Key Points

• The date property returns a numpy array of Python datetime.date objects
• Timezone information is removed from the original timestamps
• Time components (hour, minute, second) are discarded, keeping only the date
• Each element in the returned array is a standard Python datetime.date object

Conclusion

The date property of a DatetimeIndex provides an efficient way to extract date objects from timestamp data. This is useful when you only need the date portion without time or timezone information.