Polars

What is it?

Polars is a DataFrame library written in Rust with a Python API. It uses all your CPU cores and a lazy evaluation engine that optimizes the query before running it — built to be fast and memory-efficient by design.

What is it for?

• Processing large datasets (the ones pandas struggles with) on a single machine.
• Chaining declarative transformations that Polars optimizes as a whole.
• Working with streaming data that doesn't fully fit in RAM (lazy mode).

When to use it / when not

Use it when pandas falls short on speed or memory, or when you want a modern, expressive API with good performance out of the box.

Think twice if you depend on libraries that expect a pandas DataFrame, or if your dataset is small and your team's familiarity with pandas matters more than speed.

Get started in 1 minute

pip install polars

import polars as pl

# Lazy: Polars optimizes the whole plan before reading data
summary = (
    pl.scan_csv("sales.csv")
      .group_by("country")
      .agg(pl.col("amount").sum().alias("total"))
      .sort("total", descending=True)
      .collect()
)

print(summary)