Polars Cheatsheet for Data Transformation and Analysis in Python

Introduction to Polars

Polars is a Python library designed for efficient data transformation, analysis, and visualization. It was initially released by Ritchie Vink in 2020 and provides a DataFrame API for working with structured data. The library emphasizes performance and an expressive syntax for data manipulation.

Installation and Basic Usage

Users can install Polars with all its optional dependencies using `uv pip install "polars[all]"`. After installation, Polars can be imported as `pl` in Python. The library allows users to check installed versions of Polars and its dependencies using `pl.show_versions()`.

Core Data Structures and API Differences

Polars stores data in Series or DataFrames. Unlike pandas, Polars DataFrames do not include a row index, promoting immutability and method chaining over in-place modifications. Users can create Series from sequences of values and DataFrames from dictionaries or by reading files using `pl.read_*()` functions. Row indices can be added explicitly as a column using `df.with_row_index("id")` when needed.

Eager and Lazy APIs

Polars offers both eager and lazy execution APIs. The eager API executes operations immediately, while the lazy API constructs an optimized query plan before execution. This optimization automatically applies techniques like predicate pushdown (filtering early) and projection pushdown (dropping unused columns). Users can switch between a DataFrame and a LazyFrame using `.lazy()` and `.collect()` methods, respectively. The streaming engine can process datasets larger than memory by using `lf.collect(engine="streaming")`.