Polars is a blazingly fast DataFrame library, written in Rust.
The goal of Polars is to deliver fast, efficient data processing that:
- Utilizes all available cores on your machine.
- Optimizes queries to reduce unneeded work/memory allocations.
- Handles datasets much larger than your available RAM.
- Follows a consistent and predictable API.
- Adheres to a strict schema (data-types should be known before running the query).
This polars R package provides the R bindings for Polars. It can be used to process Polars DataFrames and other data structures, convert objects between Polars and R, and can be integrated with other common R packages.
To learn more, read the online documentation for this R package, and the user guide for Python/Rust Polars.
The recommended way to install this package is from the R-multiverse community repository:
Sys.setenv(NOT_CRAN = "true")
install.packages("polars", repos = "https://community.r-multiverse.org")More recent (development) version may be installed from the rpolars R-universe repository:
Sys.setenv(NOT_CRAN = "true")
install.packages('polars', repos = c("https://rpolars.r-universe.dev", "https://cloud.r-project.org"))To avoid conflicts with other packages and base R function names, many
of polars’ functions are hosted in the pl environment, and accessible
via the pl$ prefix. Most of the functions for polars objects should be
chained with the $ operator.
Additionally, the majority of the functions are intended to match the Python Polars API.
These mean that Polars queries written in Python and in R are very similar.
For example, writing the example from the user guide of Python/Rust in R:
library(neopolars) # TODO: rename to polars
# Prepare a CSV file
csv_file <- tempfile(fileext = ".csv")
write.csv(iris, csv_file, row.names = FALSE)
# Create a query plan (LazyFrame) with filtering and group aggregation
q <- pl$scan_csv(csv_file)$filter(pl$col("Sepal.Length") > 5)$group_by(
"Species",
.maintain_order = TRUE
)$agg(pl$all()$sum())
# Execute the query plan and collect the result as a Polars DataFrame
df <- q$collect()
df
#> shape: (3, 5)
#> ┌────────────┬──────────────┬─────────────┬──────────────┬─────────────┐
#> │ Species ┆ Sepal.Length ┆ Sepal.Width ┆ Petal.Length ┆ Petal.Width │
#> │ --- ┆ --- ┆ --- ┆ --- ┆ --- │
#> │ str ┆ f64 ┆ f64 ┆ f64 ┆ f64 │
#> ╞════════════╪══════════════╪═════════════╪══════════════╪═════════════╡
#> │ setosa ┆ 116.9 ┆ 81.7 ┆ 33.2 ┆ 6.1 │
#> │ versicolor ┆ 281.9 ┆ 131.8 ┆ 202.9 ┆ 63.3 │
#> │ virginica ┆ 324.5 ┆ 146.2 ┆ 273.1 ┆ 99.6 │
#> └────────────┴──────────────┴─────────────┴──────────────┴─────────────┘There are also some functions to manipulate polars objects using base R and some popular other packages.
# Subset a Polars DataFrame using the `[` operator
df[1:2, 1:2]
#> shape: (2, 2)
#> ┌────────────┬──────────────┐
#> │ Species ┆ Sepal.Length │
#> │ --- ┆ --- │
#> │ str ┆ f64 │
#> ╞════════════╪══════════════╡
#> │ setosa ┆ 116.9 │
#> │ versicolor ┆ 281.9 │
#> └────────────┴──────────────┘
# Execute a query plan and collect the result as a tibble data frame
tibble::as_tibble(q)
#> # A tibble: 3 × 5
#> Species Sepal.Length Sepal.Width Petal.Length Petal.Width
#> <chr> <dbl> <dbl> <dbl> <dbl>
#> 1 setosa 117. 81.7 33.2 6.1
#> 2 versicolor 282. 132. 203. 63.3
#> 3 virginica 324. 146. 273. 99.6The Get Started
vignette
(vignette("polars")) provides a more detailed introduction.
While one can use this package as-is, other packages build on it to provide different APIs:
- polarssql provides a polars backend for DBI and dbplyr.
- tidypolars allows one to use the tidyverse syntax while using the power of polars.
Version 0 of R Polars was previously maintained by Søren Havelund Welling.
This package is based on the Polars open source project, originally founded by Ritchie Vink and developed by many contributors.
MIT @ polars authors