Time Series Resampling (resample, asfreq)

Resampling changes the frequency of time-indexed data — turning daily readings into monthly totals (downsampling) or hourly readings into 15-minute slots (upsampling) by grouping rows into time buckets.

Learn Time Series Resampling (resample, asfreq) in our free Pandas course — a beginner-friendly interactive lesson with worked examples, a practice exercise…

Part of the free Pandas course at LearnCodingFast — hands-on lessons with examples you run in your browser, plus practice exercises and a quick quiz.

You'll learn df.resample() with .sum() , .mean() , and .ohlc() , how upsampling differs from downsampling, and when to reach for asfreq() instead.

Resampling is "groupby for time." When your index is a DatetimeIndex , calling df.resample("ME") slices the timeline into month-end buckets; you then chain an aggregation like .sum() or .mean() to collapse each bucket into one row. Going from a fine frequency (daily) to a coarse one (monthly) is called downsampling .

The rule argument is a frequency alias . The common ones are "D" (day), "W" (week), "ME" (month-end), "QE" (quarter-end), and "YE" (year-end). You can also prefix a number, e.g. "2D" for two-day windows.

Each bucket can be summarised in many ways. .ohlc() is the classic finance summary that returns four columns per bucket — open (first value), high (max), low (min), and close (last). When you want several statistics at once, pass a list to .agg([...]) .

Upsampling goes the other direction — from coarse to fine, e.g. daily to 12-hourly. Because the new timestamps did not exist in the original data, pandas fills them with NaN . You then decide how to fill: forward-fill with .ffill() , back-fill with .bfill() , or leave the holes for interpolation (the next lesson).

asfreq() is the lightweight cousin of resample. It does not bucket or aggregate — it simply lays a new fixed-frequency index over the data and keeps whatever value sits exactly on each new timestamp, inserting NaN elsewhere. Use it when the frequency is already regular and you just want to change its label rate.

❌ "Invalid frequency: M ... use 'ME' instead"

The old single-letter month alias was removed in modern pandas:

❌ "Only valid with DatetimeIndex" / TypeError

Calling resample when the dates are still a plain column:

You have website hits recorded every 6 hours. Downsample to daily numbers two ways.

Lesson complete — you can reshape time itself!

You can downsample with resample("ME"/"W"/"D") plus an aggregation, summarise buckets with .ohlc() and .agg([...]) , upsample with fills, and re-stamp a regular series with asfreq() .

🚀 Up next: Interpolation & Filling Gaps — smart ways to fill those upsampled NaNs.

Practice quiz

• Browse all free courses
• Cheat sheets & quizzes
• Coding dictionary