Python Pandas - How to Round the DateTimeIndex with milliseconds frequency

To round the DateTimeIndex with milliseconds frequency, use the DateTimeIndex.round() method. For milliseconds frequency, use the freq parameter with value 'ms'.

Creating a DateTimeIndex

First, import pandas and create a DateTimeIndex with nanosecond precision ?

import pandas as pd

# Create DatetimeIndex with period 5 and frequency as 28 seconds
# timezone is Australia/Adelaide
datetimeindex = pd.date_range('2021-09-29 07:20:32.261811624', periods=5,
                             tz='Australia/Adelaide', freq='28s')

print("Original DateTimeIndex:")
print(datetimeindex)

Original DateTimeIndex:
DatetimeIndex(['2021-09-29 07:20:32.261811624+09:30',
               '2021-09-29 07:21:00.261811624+09:30',
               '2021-09-29 07:21:28.261811624+09:30',
               '2021-09-29 07:21:56.261811624+09:30',
               '2021-09-29 07:22:24.261811624+09:30'],
              dtype='datetime64[ns, Australia/Adelaide]', freq='28S')

Rounding to Milliseconds Frequency

Use the round() method with freq='ms' to round timestamps to the nearest millisecond ?

import pandas as pd

# Create DatetimeIndex with nanosecond precision
datetimeindex = pd.date_range('2021-09-29 07:20:32.261811624', periods=5,
                             tz='Australia/Adelaide', freq='28s')

# Round operation to milliseconds frequency
rounded_index = datetimeindex.round(freq='ms')

print("Rounded to milliseconds:")
print(rounded_index)

Rounded to milliseconds:
DatetimeIndex(['2021-09-29 07:20:32.262000+09:30',
               '2021-09-29 07:21:00.262000+09:30',
               '2021-09-29 07:21:28.262000+09:30',
               '2021-09-29 07:21:56.262000+09:30',
               '2021-09-29 07:22:24.262000+09:30'],
              dtype='datetime64[ns, Australia/Adelaide]', freq=None)

Complete Example

Here's a comprehensive example showing the original DateTimeIndex, its frequency, and the rounded result ?

import pandas as pd

# DatetimeIndex with period 5 and frequency as 28 seconds
# timezone is Australia/Adelaide
datetimeindex = pd.date_range('2021-09-29 07:20:32.261811624', periods=5,
                             tz='Australia/Adelaide', freq='28s')

# Display DateTimeIndex
print("DateTimeIndex...")
print(datetimeindex)

# Display DateTimeIndex frequency
print("\nDateTimeIndex frequency...")
print(datetimeindex.freq)

# Round operation on DateTimeIndex date with milliseconds frequency
# For milliseconds frequency, we have used 'ms'
print("\nPerforming round operation with milliseconds frequency...")
print(datetimeindex.round(freq='ms'))

DateTimeIndex...
DatetimeIndex(['2021-09-29 07:20:32.261811624+09:30',
               '2021-09-29 07:21:00.261811624+09:30',
               '2021-09-29 07:21:28.261811624+09:30',
               '2021-09-29 07:21:56.261811624+09:30',
               '2021-09-29 07:22:24.261811624+09:30'],
              dtype='datetime64[ns, Australia/Adelaide]', freq='28S')

DateTimeIndex frequency...
<28 * Seconds>

Performing round operation with milliseconds frequency...
DatetimeIndex(['2021-09-29 07:20:32.262000+09:30',
               '2021-09-29 07:21:00.262000+09:30',
               '2021-09-29 07:21:28.262000+09:30',
               '2021-09-29 07:21:56.262000+09:30',
               '2021-09-29 07:22:24.262000+09:30'],
              dtype='datetime64[ns, Australia/Adelaide]', freq=None)

Key Points

• The original timestamps have nanosecond precision (261811624 nanoseconds)
• Rounding to milliseconds converts this to 262000 microseconds (262 milliseconds)
• The freq parameter becomes None after rounding as the regular frequency is lost
• Timezone information is preserved during the rounding operation

Conclusion

Use DateTimeIndex.round(freq='ms') to round timestamps to millisecond precision. This removes microsecond and nanosecond components while preserving timezone information.