Skip to content
Laviyx Learn

Try

Python for Data course menu

Dates and Time Series in pandas: to_datetime, dt and resample

Work with dates in pandas: parse dates, extract year, month and weekday with .dt, date arithmetic with Timedelta, date ranges, filtering by date, a DatetimeIndex, and resampling daily data to weekly or monthly totals.

Intermediate9 min readUpdated
On this page

Sales per month, sign-ups per weekday, days since last purchase: time is part of most business questions. pandas has a full set of tools for dates once a column has the datetime type.

Parsing dates#

Python
import pandas as pd

orders = pd.DataFrame({
    "ordered": ["2026-01-05", "2026-01-28", "2026-02-11", "2026-03-02", "2026-03-30"],
    "amount": [250, 120, 560, 90, 310],
})
orders["ordered"] = pd.to_datetime(orders["ordered"])
print(orders.dtypes)
Output
ordered    datetime64[ns]
amount              int64
dtype: object

ISO dates (YYYY-MM-DD) parse without a format. For anything else, pass format= (see Data Types).

The dt accessor#

.dt gives every date part as a column:

Python
import pandas as pd

orders = pd.DataFrame({
    "ordered": pd.to_datetime(["2026-01-05", "2026-01-28", "2026-02-11", "2026-03-02", "2026-03-30"]),
    "amount": [250, 120, 560, 90, 310],
})
orders["year"] = orders["ordered"].dt.year
orders["month"] = orders["ordered"].dt.month
orders["weekday"] = orders["ordered"].dt.day_name()
orders["quarter"] = orders["ordered"].dt.quarter
orders["month_label"] = orders["ordered"].dt.strftime("%b %Y")
print(orders)
print(orders.groupby("month_label", sort=False)["amount"].sum())
Output
     ordered  amount  year  month    weekday  quarter month_label
0 2026-01-05     250  2026      1     Monday        1    Jan 2026
1 2026-01-28     120  2026      1  Wednesday        1    Jan 2026
2 2026-02-11     560  2026      2  Wednesday        1    Feb 2026
3 2026-03-02      90  2026      3     Monday        1    Mar 2026
4 2026-03-30     310  2026      3     Monday        1    Mar 2026
month_label
Jan 2026    370
Feb 2026    560
Mar 2026    400
Name: amount, dtype: int64

strftime formats dates as text with the same codes used for parsing.

Date arithmetic#

Subtracting dates gives a Timedelta; adding a Timedelta or DateOffset moves dates:

Python
import pandas as pd

ordered = pd.to_datetime(pd.Series(["2026-01-05", "2026-02-11", "2026-03-30"]))
today = pd.Timestamp("2026-04-01")

print((today - ordered).dt.days.tolist())                  # days since each order
print((ordered + pd.Timedelta(days=7)).dt.date.tolist())    # due a week later
print((ordered + pd.DateOffset(months=1)).dt.date.tolist()) # same day next month
print(ordered.dt.to_period("M").tolist())
Output
[86, 49, 2]
[datetime.date(2026, 1, 12), datetime.date(2026, 2, 18), datetime.date(2026, 4, 6)]
[datetime.date(2026, 2, 5), datetime.date(2026, 3, 11), datetime.date(2026, 4, 30)]
[Period('2026-01', 'M'), Period('2026-02', 'M'), Period('2026-03', 'M')]

DateOffset(months=1) respects calendar months, which a fixed number of days cannot. to_period("M") labels each date with its month.

Date ranges#

Python
import pandas as pd

print(pd.date_range("2026-03-01", periods=5, freq="D").date.tolist())
print(pd.date_range("2026-01-01", "2026-06-30", freq="MS").strftime("%b").tolist())
print(pd.date_range("2026-03-02", periods=3, freq="W-MON").date.tolist())
Output
[datetime.date(2026, 3, 1), datetime.date(2026, 3, 2), datetime.date(2026, 3, 3), datetime.date(2026, 3, 4), datetime.date(2026, 3, 5)]
['Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun']
[datetime.date(2026, 3, 2), datetime.date(2026, 3, 9), datetime.date(2026, 3, 16)]

Frequency codes: D day, W-MON weekly on Mondays, MS month start, ME month end, QE quarter end, h hour. (Older pandas used M and Q; recent versions want ME and QE for month and quarter ends.)

A DatetimeIndex and filtering#

Setting the date as the index makes time-based selection easy:

Python
import numpy as np
import pandas as pd

rng = np.random.default_rng(3)
daily = pd.Series(rng.integers(80, 160, size=90),
                  index=pd.date_range("2026-01-01", periods=90, freq="D"), name="visitors")
print(daily.head(3))
print(daily.loc["2026-02"].sum())                       # all of February
print(daily.loc["2026-03-10":"2026-03-12"])              # a date range (inclusive)
print(daily[daily.index.dayofweek >= 5].mean().round(1))  # weekends only
Output
2026-01-01    144
2026-01-02     86
2026-01-03     94
Freq: D, Name: visitors, dtype: int64
3331
2026-03-10    154
2026-03-11     96
2026-03-12    128
Freq: D, Name: visitors, dtype: int64
119.1

Resampling#

resample is groupby for time: it groups by calendar periods. Daily data can become weekly or monthly totals in one line:

Python
import numpy as np
import pandas as pd

rng = np.random.default_rng(3)
daily = pd.Series(rng.integers(80, 160, size=90),
                  index=pd.date_range("2026-01-01", periods=90, freq="D"), name="visitors")
monthly = daily.resample("ME").sum()
monthly.index = monthly.index.strftime("%b")
print(monthly)
print(daily.resample("W-SUN").mean().round(1).head(3))
Output
Jan    3566
Feb    3331
Mar    3784
Name: visitors, dtype: int64
2026-01-04    105.5
2026-01-11    112.7
2026-01-18    118.1
Freq: W-SUN, Name: visitors, dtype: float64

Grouping without a DatetimeIndex#

If the date is a regular column, group by a period instead:

Python
import pandas as pd

orders = pd.DataFrame({
    "ordered": pd.to_datetime(["2026-01-05", "2026-01-28", "2026-02-11", "2026-03-02", "2026-03-30"]),
    "amount": [250, 120, 560, 90, 310],
})
print(orders.groupby(orders["ordered"].dt.to_period("M"))["amount"].agg(["sum", "count"]))
print(orders.groupby(pd.Grouper(key="ordered", freq="QE"))["amount"].sum())
Output
         sum  count
ordered
2026-01  370      2
2026-02  560      1
2026-03  400      2
ordered
2026-03-31    1330
Freq: QE-DEC, Name: amount, dtype: int64

Common mistakes#

Practice#

Practice exercise

Sign-up dates are ["2026-01-03", "2026-01-15", "2026-02-01", "2026-02-14", "2026-02-20", "2026-04-02"]. Print the number of sign-ups per month including months with none (January to April), and the weekday on which most sign-ups happened.

Show hint

Make a Series of 1s indexed by the dates and resample("ME").sum(). For weekdays, dt.day_name().value_counts().idxmax().

Show solution
Python
import pandas as pd

dates = pd.to_datetime(pd.Series(["2026-01-03", "2026-01-15", "2026-02-01", "2026-02-14", "2026-02-20", "2026-04-02"]))
per_month = pd.Series(1, index=dates).resample("ME").sum()
per_month.index = per_month.index.strftime("%b")
print(per_month)
print(dates.dt.day_name().value_counts().idxmax())
Output
Jan    2
Feb    3
Mar    0
Apr    1
dtype: int64
Saturday

March appears with 0 because resample creates every month in the range.

Key takeaway#

Next: Rolling Windows and Cumulative Values.

Finished this lesson?

Related lessons

Python for Datapandas Data Types: astype, to_numeric, to_datetime and categoryFix column types in pandas: check dtypes, convert with astype, clean numbers stored as text with to_numeric, parse dates with to_datetime, use the category type to save memory, and nullable integer and string types.Intermediate9 min readPython for DataRolling Averages, Cumulative Sums and Percent Change in pandasSmooth and compare time series in pandas: rolling means and sums, centred and time-based windows, expanding windows, cumulative sums, shift for previous values, and pct_change for growth rates.Intermediate9 min readPythonPython Date and Time: datetime, timedelta, strftime and Time ZonesLearn Python's datetime module: create dates and times, get today's date, format with strftime, parse strings with strptime, add days with timedelta, calculate differences and ages, and handle time zones.Intermediate7 min read