Implementation:Pola rs Polars Upsample and Interpolate

Knowledge Sources	Polars Polars Docs
Domains	Data Engineering, Time Series
Last Updated	2026-02-09 10:00 GMT

Overview

Concrete APIs for upsampling a Polars DataFrame to a higher temporal frequency and filling the resulting null values using forward fill, interpolation, or other strategies.

Description

DataFrame.upsample(time_column, every) inserts new rows at the specified frequency interval, producing a DataFrame with evenly-spaced timestamps. The non-index columns in newly inserted rows contain null values. These nulls are then filled using DataFrame.fill_null(strategy) for strategy-based filling or DataFrame.interpolate() for linear interpolation between adjacent known values.

The upsample method requires the temporal column to be sorted. An optional group_by parameter supports upsampling panel data where each entity is upsampled independently.

Usage

Use these APIs whenever you need to:

• Increase the temporal frequency of a time series (e.g., 30-minute to 15-minute intervals).
• Fill temporal gaps with forward-filled or interpolated values.
• Align multiple time series to a common temporal grid.

Code Reference

Source Location

• Repository: Polars
• File: docs/source/src/python/user-guide/transformations/time-series/resampling.py (lines 1-36)

Signature

DataFrame.upsample(
    time_column: str,
    *,
    every: str | timedelta,
    group_by: str | Sequence[str] | None = None,
    maintain_order: bool = False,
) -> DataFrame

DataFrame.fill_null(
    value: Any | Expr | None = None,
    strategy: str | None = None,
    limit: int | None = None,
    matches_supertype: bool = True,
) -> DataFrame

DataFrame.interpolate() -> DataFrame

Import

import polars as pl

I/O Contract

Inputs

Name	Type	Required	Description
time_column	str	Yes	Name of the sorted temporal column to upsample
every	timedelta	Yes	Target frequency as a duration string (e.g., "15m", "1h", "1d")
group_by	Sequence[str] | None	No	Grouping columns for panel data upsampling
strategy	None	No (for fill_null)	Fill strategy: "forward", "backward", "mean", "zero", "one", "max", "min"
limit	None	No (for fill_null)	Maximum number of consecutive nulls to fill

Outputs

Name	Type	Description
result (upsample)	pl.DataFrame	DataFrame with new rows inserted at the target frequency; non-index columns contain nulls at new time points
result (fill_null)	pl.DataFrame	DataFrame with nulls replaced according to the specified strategy
result (interpolate)	pl.DataFrame	DataFrame with nulls filled via linear interpolation between adjacent known values

Usage Examples

Upsample with Forward Fill

import polars as pl

# Upsample from 30-min to 15-min intervals
df_upsampled = df.upsample(time_column="time", every="15m")

# Fill nulls with forward fill
df_filled = df_upsampled.fill_null(strategy="forward")
print(df_filled)

Upsample with Interpolation

import polars as pl

# Upsample from 30-min to 15-min intervals
df_upsampled = df.upsample(time_column="time", every="15m")

# Fill nulls using linear interpolation
df_interpolated = df_upsampled.interpolate()
print(df_interpolated)

Upsample Panel Data

import polars as pl

# Upsample each sensor independently
df_upsampled = df.upsample(
    time_column="timestamp",
    every="1h",
    group_by="sensor_id",
)
df_filled = df_upsampled.fill_null(strategy="forward")

Related Pages

Implements Principle

• Principle:Pola_rs_Polars_Time_Series_Resampling