Skip to main content
The datatable.time module provides functions for working with temporal columns (date32 and time64). You can extract individual components (year, month, day, etc.) from a date or timestamp column, compute derived values such as day-of-week, or construct new temporal columns from their components. 
import datatable as dt
from datatable import f

DT = dt.Frame(ts=["2023-01-15", "2024-07-04", "2025-12-31"], stype=dt.Type.date32)

# Extract year, month, and day as separate columns
DT[:, {
    "year":  dt.time.year(f.ts),
    "month": dt.time.month(f.ts),
    "day":   dt.time.day(f.ts),
}]
All dt.time extractor functions accept FExpr[date32] or FExpr[time64] columns and return integer FExpr results. They operate lazily and must be used inside a DT[rows, cols] expression.

​
Component extraction functions

​
Date components

FunctionInput typesReturnsDescription
dt.time.year(date)date32, time64int32Year component.
dt.time.month(date)date32, time64int32Month component (1–12).
dt.time.day(date)date32, time64int32Day-of-month component (1–31).
dt.time.day_of_week(date)date32, time64int32Day of week (Monday = 1, Sunday = 7).

​
Time components

FunctionInput typesReturnsDescription
dt.time.hour(ts)time64int32Hour component (0–23).
dt.time.minute(ts)time64int32Minute component (0–59).
dt.time.second(ts)time64int32Second component (0–59).
dt.time.nanosecond(ts)time64int32Nanosecond component (0–999,999,999).

​
Construction functions

​
dt.time.ymd(year, month, day)

Create a date32 column from integer year, month, and day component columns. Invalid calendar combinations (e.g., February 30) produce NA. Parameters
​
year
FExpr[int]
Year values.
​
month
FExpr[int]
Month values (1–12). Values outside this range produce NA.
​
day
FExpr[int]
Day values (1 to the last day of the given month). Invalid day/month combinations produce NA.
Returns FExpr[date32].

​
dt.time.ymdt(year, month, day, hour, minute, second, nanosecond=0, *, date=None)

Create a time64 column from component columns. Invalid date combinations produce NA; time components have no range checks (e.g., you can pass second=3600 instead of hour=1). You can replace the year, month, day triple with a single date parameter of type date32. Parameters
​
year
FExpr[int]
Year values.
​
month
FExpr[int]
Month values (1–12).
​
day
FExpr[int]
Day values (1 to the last day of the given month).
​
hour
FExpr[int]
Hour values.
​
minute
FExpr[int]
Minute values.
​
second
FExpr[int]
Second values.
​
nanosecond
FExpr[int]
default:"0"
Nanosecond values. Optional.
​
date
FExpr[date32]
Alternative to the year/month/day triple. Cannot be combined with those parameters.
Returns FExpr[time64].

​
Examples

​
Extract date parts from a date32 column

import datatable as dt
from datatable import f

# Build a frame from integer-encoded date32 values
DT = dt.Frame([1, 1000, 100000], stype=dt.Type.date32)

DT[:, {
    "date":  f[0],
    "year":  dt.time.year(f[0]),
    "month": dt.time.month(f[0]),
    "day":   dt.time.day(f[0]),
}]
#    | date         year  month    day
#    | date32      int32  int32  int32
# -- + ----------  -----  -----  -----
#  0 | 1970-01-02   1970      1      2
#  1 | 1972-09-27   1972      9     27
#  2 | 2243-10-17   2243     10     17

​
Extract time components from a time64 column

import datatable as dt
from datatable import f

DT = dt.Frame(ts=["2024-06-15T14:30:45.123456", "2025-01-01T00:00:00"],
              stype=dt.Type.time64)

DT[:, {
    "hour":   dt.time.hour(f.ts),
    "minute": dt.time.minute(f.ts),
    "second": dt.time.second(f.ts),
}]

​
Construct a date32 column from parts

import datatable as dt
from datatable import f

DT = dt.Frame(y=[2005, 2010, 2015], m=[2, 3, 7])

# February 30 is invalid → NA; others succeed
DT[:, dt.time.ymd(f.y, f.m, 30)]
#    | C0
#    | date32
# -- + ----------
#  0 | NA
#  1 | 2010-03-30
#  2 | 2015-07-30

​
Filter rows by day of week

import datatable as dt
from datatable import f

DT = dt.Frame(event=["A", "B", "C", "D", "E"],
              date=["2024-06-10", "2024-06-11", "2024-06-12",
                    "2024-06-13", "2024-06-14"],
              stype={"date": dt.Type.date32})

# Keep only weekdays (Monday=1 … Friday=5)
weekdays = DT[dt.time.day_of_week(f.date) <= 5, :]

Build docs developers (and LLMs) love

Get started for freeTalk to us