Lightnews — Scholar-powered news

Etienne Bacher @etiennebacher.bsky.social · 1d

I might have something for you in a few weeks

Reposted by Etienne Bacher

Vincent Arel-Bundock @vincentab.bsky.social · 9d

{tinytable} 0.14.0 for #RStats makes it super easy to draw tables in html, tex, docx, typ, md & png.

There are only a few functions to learn, but don't be fooled! Small 📦s can still be powerful.

Check out the new gallery page for fun case studies.

vincentarelbundock.github.io/tinytable/vi...

1 39 130

Etienne Bacher @etiennebacher.bsky.social · 13d

I remember seeing {debugme} being used in several r-lib packages as well (I've never used it however):

github.com/r-lib/debugme

GitHub - r-lib/debugme: Easy and efficient debugging for R packages

Easy and efficient debugging for R packages. Contribute to r-lib/debugme development by creating an account on GitHub.

github.com

1

Etienne Bacher @etiennebacher.bsky.social · Jul 24

Other features include:
* improved support for lubridate functions
* experimental support for fallback to R when a function cannot be
translated to polars
* performance improvements when handling thousands of columns

and more.

Full changelog: www.tidypolars.etiennebacher.com/news/

(2/2)

Changelog

www.tidypolars.etiennebacher.com

2 2

Etienne Bacher @etiennebacher.bsky.social · Jul 24

#rstats tidypolars 0.14.0 is now available!

tidypolars provides the tidyverse syntax while using polars for efficient computations in the background.

This release brings compatibility with polars 1.0.0, released earlier
this week, and several features and bug fixes.

(1/2)

2 6 22

Etienne Bacher @etiennebacher.bsky.social · Jul 23

Yes, there's a learning curve for sure.

I have also developed tidypolars to keep using tidyverse code while using polars in the background: tidypolars.etiennebacher.com

This takes care of most dataframe-level functions, and there's a number of other functions supported.

More Efficient Tidyverse Code, Using Polars in the Background

Polars is a cross-language tool for manipulating very large data. However, one drawback is that the R implementation has a syntax that will look odd to many R users who are not used to Python syntax. ...

tidypolars.etiennebacher.com

1 2 7

Etienne Bacher @etiennebacher.bsky.social · Jul 23

Would love to but looks unlikely. The upstream polars library uses very recent Rust versions which are not available on CRAN machines, and they update the rust version they use quite frequently.

R-universe / R-multiverse is the best way to get polars binaries for now.

6

Etienne Bacher @etiennebacher.bsky.social · Jul 23

polars can be used via webR, thanks to the work of @gws.phd both on webR and on the Rust component of polars.

It also uses savvy, developed by @yutannihilation-en.bsky.social‬, to connect Rust and R.

2 6

Etienne Bacher @etiennebacher.bsky.social · Jul 23

polars 1.0.0 contains many breaking changes because internals have changed or because we better match the function and argument names provided in python polars.

We encourage you to refer to the python polars documentation if you use some functions or arguments that don't exist anymore.

1 1 1

Etienne Bacher @etiennebacher.bsky.social · Jul 23

The way to handle custom R functions has been revamped and no longer relies on map_elements() or map_batches(). Instead, if your code cannot be implemented using polars syntax, we encourage the use of purrr and mirai.

A new vignette explains this in more details: pola-rs.github.io/r-polars/vig...

Using custom functions - Polars R Package

pola-rs.github.io

1 1 1

Etienne Bacher @etiennebacher.bsky.social · Jul 23

Catching up with the python implementation means that we now have access to the new streaming engine with collect(engine = "streaming").

This is in many cases faster than the default engine and allows processing data that is larger than RAM. It will also continue to improve in future releases.

Some code (joining some flight data with some weather data) that runs using the standard engine, takes about 11s to finish.

Some code (joining some flight data with some weather data) that runs using the streaming engine, takes about 6s to finish.

1 1 1

Etienne Bacher @etiennebacher.bsky.social · Jul 23

polars now uses rlang to take advantages of dynamic dots, better checks, and clearer error messages.

Compare for instance the old (polars0) and new (polars) errors when a column name doesn't exist:

A picture of code that shows an error message with the old version of polars.

> library(polars0)
> pl$DataFrame(x = "a")$select(pl$col("a")$str$pad_end(5, "multiple_chars"))
Error:
! Execution halted with the following contexts
0: In R: in $select()
0: During function call [pl$DataFrame(x = "a")$select(pl$col("a")$str$pad_end(5, "multiple_chars"))]
1: The argument [fillchar] caused an error
2: Expected a value of type [char]
3: Got value [Rvalue: ["multiple_chars"], Rsexp: Strings, Rclass: ["character"]]

A picture showing the error message with the new version of polars:

> library(polars)
> pl$DataFrame(x = "a")$select(pl$col("a")$str$pad_end(5, "multiple_chars"))
Error:
! Evaluation failed in `$select()`.
Caused by error:
! Evaluation failed in `$select()`.
Caused by error in `pl$col("a")$str$pad_end()`:
! Evaluation failed in `$pad_end()`.
Caused by error:
! Expected a string with one character only, currently has 14 (from "multiple_chars").

1 1 1

Etienne Bacher @etiennebacher.bsky.social · Jul 23

#rstats polars 1.0.0 is now available!

polars provides bindings to the eponymous Rust library, a very efficient dataframe library.

This is the outcome of a 1-year long rewrite led by @eitsupi to refactor the internals and match the API of Python polars.

pola-rs.github.io/r-polars/

Details ⬇️

Polars R Package

pola-rs.github.io

4 12 36

Etienne Bacher @etiennebacher.bsky.social · Jul 7

You could also be interested in tidypolars, especially for (but not limited to) parquet datasets.

(Disclaimer: I'm the author)

Same as duckplyr, that depends on whether the functions you use are translated.

Major drawback is that it's not on CRAN: tidypolars.etiennebacher.com

More Efficient Tidyverse Code, Using Polars in the Background

Polars is a cross-language tool for manipulating very large data. However, one drawback is that the R implementation has a syntax that will look odd to many R users who are not used to Python syntax. ...

tidypolars.etiennebacher.com

1

Etienne Bacher @etiennebacher.bsky.social · Jul 7

Wow that's surprisingly low to me

3

Etienne Bacher @etiennebacher.bsky.social · Jul 7

Is this about dependencies mentioned in DESCRIPTION or also recursive ones?

1 1

Etienne Bacher @etiennebacher.bsky.social · Jun 30

fix_*() functions now have an argument `interactive` to show suggested changes in a Shiny app before applying them (see the gif below). This is useful to check the changes instead of modifying several files at once.

(3/3)

A gif that shows some lints in two R files (e.g. any(is.na(x))). Then it shows that flir::fix(c("foo1.R", "foo2.R"), interactive = TRUE) opens a Shiny app in the viewer so that the user can decide whether the proposed changes should be accepted or not.

1 1

Etienne Bacher @etiennebacher.bsky.social · Jun 30

Package developers can now provide customized rules for {flir} users, for instance to detect and replace deprecated or superseded functions.

This is detailed in the following vignette and is still an early feature, so feedback is much appreciated: flir.etiennebacher.com/articles/sha...

(2/3)

Sharing rules across packages

flir.etiennebacher.com

1 1 2

Etienne Bacher @etiennebacher.bsky.social · Jun 30

#rstats {flir} 0.5.0 is now available!

{flir} can detect and rewrite code patterns.

This is the 1st CRAN release and has two nice features:
- share custom rules across projects
- show changes in Shiny app before applying them

flir.etiennebacher.com/news/#flir-050

More details in the 🧵 (1/3)

Changelog

flir.etiennebacher.com

1 1 5

Reposted by Etienne Bacher

Laurent Bergé @lrberge.bsky.social · Jun 20

Just gave a presentation a the REGIS summer school we organised in Bordeaux.

Algorithms for empirical research

Slides and material here: github.com/lrberge/2025...

The conclusion is that, even in the age of AI, using your brain is still relevant!

#RStats #EconSky

2 5 12

Etienne Bacher @etiennebacher.bsky.social · Jun 16

Nope but someone writes rv, that I haven't tried: github.com/A2-ai/rv

1

Etienne Bacher @etiennebacher.bsky.social · Jun 15

Used to agree but uv really changes that. It's very good

1

Etienne Bacher @etiennebacher.bsky.social · Jun 11

...it's better to let the students "suffer" with projects that dont have version control. To me, it's easier to see the benefits of this tool after you have spent some time without it, so it's clearer what gap it fills in the workflow.

So in an intro course I would teach projects but not git (2/2)

1 1 1

Etienne Bacher @etiennebacher.bsky.social · Jun 11

Good question, I think it depends on the tool you teach. For instance, Rstudio projects shouldn't be too hard to understand (despite the caveat about tech savvy you mentioned), so I try to introduce this early.

Other tools are harder to understand, such as git. In this case, I think ... (1/2)

1 1 1

Etienne Bacher @etiennebacher.bsky.social · Jun 11

Totally agree with you, my point was just that regardless whether you teach them rstudio or positron, you will have to teach them how to properly use a project setup.

There are reasons to prefer rstudio over positron, I just don't think Rstudio projects is the main one.

1 2