polars is a DataFrame library written in Rust and with bindings in
several languages, including R. I won’t
argue here for the interest of using polars, there are already a lot
of resources on its website.
The R package polars was made to mimic very closely the original
Python syntax, which is quite verbose. While this makes it easy to read,
it is yet another syntax to learn for R users that are accustomed so
far to either base R, data.table or the tidyverse.
The objective of tidypolars is to provide the power and speed of
polars while keeping the tidyverse syntax.
tidypolars is built on polars, which is not available on CRAN. This
means that tidypolars also can’t be on CRAN. However, you can install
it from R-universe.
install.packages(
'tidypolars',
repos = c('https://etiennebacher.r-universe.dev', getOption("repos"))
)install.packages(
'tidypolars',
repos = c('https://etiennebacher.r-universe.dev/bin/linux/jammy/4.3', getOption("repos"))
)Suppose that you already have some code that uses dplyr:
library(dplyr, warn.conflicts = FALSE)
iris |>
select(starts_with(c("Sep", "Pet"))) |>
mutate(
petal_type = ifelse((Petal.Length / Petal.Width) > 3, "long", "large")
) |>
filter(between(Sepal.Length, 4.5, 5.5)) |>
head()
#> Sepal.Length Sepal.Width Petal.Length Petal.Width petal_type
#> 1 5.1 3.5 1.4 0.2 long
#> 2 4.9 3.0 1.4 0.2 long
#> 3 4.7 3.2 1.3 0.2 long
#> 4 4.6 3.1 1.5 0.2 long
#> 5 5.0 3.6 1.4 0.2 long
#> 6 5.4 3.9 1.7 0.4 longWith tidypolars, you can provide a Polars DataFrame or LazyFrame
and keep the exact same code:
library(tidypolars)
#> Registered S3 method overwritten by 'tidypolars':
#> method from
#> print.RPolarsDataFrame polars
iris |>
as_polars_df() |>
select(starts_with(c("Sep", "Pet"))) |>
mutate(
petal_type = ifelse((Petal.Length / Petal.Width) > 3, "long", "large")
) |>
filter(between(Sepal.Length, 4.5, 5.5)) |>
head()
#> shape: (6, 5)
#> ┌──────────────┬─────────────┬──────────────┬─────────────┬────────────┐
#> │ Sepal.Length ┆ Sepal.Width ┆ Petal.Length ┆ Petal.Width ┆ petal_type │
#> │ --- ┆ --- ┆ --- ┆ --- ┆ --- │
#> │ f64 ┆ f64 ┆ f64 ┆ f64 ┆ str │
#> ╞══════════════╪═════════════╪══════════════╪═════════════╪════════════╡
#> │ 5.1 ┆ 3.5 ┆ 1.4 ┆ 0.2 ┆ long │
#> │ 4.9 ┆ 3.0 ┆ 1.4 ┆ 0.2 ┆ long │
#> │ 4.7 ┆ 3.2 ┆ 1.3 ┆ 0.2 ┆ long │
#> │ 4.6 ┆ 3.1 ┆ 1.5 ┆ 0.2 ┆ long │
#> │ 5.0 ┆ 3.6 ┆ 1.4 ┆ 0.2 ┆ long │
#> │ 5.4 ┆ 3.9 ┆ 1.7 ┆ 0.4 ┆ long │
#> └──────────────┴─────────────┴──────────────┴─────────────┴────────────┘If you’re used to the tidyverse functions and syntax, this will feel
much easier to read than the pure polars syntax:
library(polars)
# polars syntax
pl$DataFrame(iris)$
select(c("Sepal.Length", "Sepal.Width", "Petal.Length", "Petal.Width"))$
with_columns(
pl$when(
(pl$col("Petal.Length") / pl$col("Petal.Width") > 3)
)$then(pl$lit("long"))$
otherwise(pl$lit("large"))$
alias("petal_type")
)$
filter(pl$col("Sepal.Length")$is_between(4.5, 5.5))$
head(6)
#> shape: (6, 5)
#> ┌──────────────┬─────────────┬──────────────┬─────────────┬────────────┐
#> │ Sepal.Length ┆ Sepal.Width ┆ Petal.Length ┆ Petal.Width ┆ petal_type │
#> │ --- ┆ --- ┆ --- ┆ --- ┆ --- │
#> │ f64 ┆ f64 ┆ f64 ┆ f64 ┆ str │
#> ╞══════════════╪═════════════╪══════════════╪═════════════╪════════════╡
#> │ 5.1 ┆ 3.5 ┆ 1.4 ┆ 0.2 ┆ long │
#> │ 4.9 ┆ 3.0 ┆ 1.4 ┆ 0.2 ┆ long │
#> │ 4.7 ┆ 3.2 ┆ 1.3 ┆ 0.2 ┆ long │
#> │ 4.6 ┆ 3.1 ┆ 1.5 ┆ 0.2 ┆ long │
#> │ 5.0 ┆ 3.6 ┆ 1.4 ┆ 0.2 ┆ long │
#> │ 5.4 ┆ 3.9 ┆ 1.7 ┆ 0.4 ┆ long │
#> └──────────────┴─────────────┴──────────────┴─────────────┴────────────┘Since most of the work is rewriting tidyverse code into polars
syntax, tidypolars and polars have very similar performance.
Click to see a small benchmark
For more serious benchmarks about polars, take a look at DuckDB
benchmarks.
library(collapse, warn.conflicts = FALSE)
#> collapse 2.0.7, see ?`collapse-package` or ?`collapse-documentation`
large_iris <- data.table::rbindlist(rep(list(iris), 100000))
large_iris_pl <- as_polars_lf(large_iris)
format(nrow(large_iris), big.mark = ",")
#> [1] "15,000,000"
bench::mark(
polars = {
large_iris_pl$
select(c("Sepal.Length", "Sepal.Width", "Petal.Length", "Petal.Width"))$
with_columns(
pl$when(
(pl$col("Petal.Length") / pl$col("Petal.Width") > 3)
)$then(pl$lit("long"))$
otherwise(pl$lit("large"))$
alias("petal_type")
)$
filter(pl$col("Sepal.Length")$is_between(4.5, 5.5))$
collect()
},
tidypolars = {
large_iris_pl |>
select(starts_with(c("Sep", "Pet"))) |>
mutate(
petal_type = ifelse((Petal.Length / Petal.Width) > 3, "long", "large")
) |>
filter(between(Sepal.Length, 4.5, 5.5)) |>
collect()
},
dplyr = {
large_iris |>
select(starts_with(c("Sep", "Pet"))) |>
mutate(
petal_type = ifelse((Petal.Length / Petal.Width) > 3, "long", "large")
) |>
filter(between(Sepal.Length, 4.5, 5.5))
},
collapse = {
large_iris |>
fselect(c("Sepal.Length", "Sepal.Width", "Petal.Length", "Petal.Width")) |>
fmutate(
petal_type = data.table::fifelse((Petal.Length / Petal.Width) > 3, "long", "large")
) |>
fsubset(Sepal.Length >= 4.5 & Sepal.Length <= 5.5)
},
check = FALSE,
iterations = 40
)
#> Warning: Some expressions had a GC in every iteration; so filtering is
#> disabled.
#> # A tibble: 4 × 6
#> expression min median `itr/sec` mem_alloc `gc/sec`
#> <bch:expr> <bch:tm> <bch:tm> <dbl> <bch:byt> <dbl>
#> 1 polars 161.36ms 205.42ms 4.22 27.86KB 0
#> 2 tidypolars 197.96ms 268.42ms 3.13 290.98KB 0.235
#> 3 dplyr 5.14s 5.78s 0.168 1.79GB 0.491
#> 4 collapse 490.3ms 618.3ms 1.56 745.96MB 1.37
# NOTE: do NOT take the "mem_alloc" results into account.
# `bench::mark()` doesn't report the accurate memory usage for packages calling
# Rust code.Did you find some errors in the documentation? Do you want tidypolars
to support more functions?
Take a look at the contributing guide!