mlflow

MLflow docs require manually downloading win-quality.csv.

r2d3

The r2d2 package docs require downloading flare.csv.

readr

The readr package defines readr_example() to avoid downloads.

R4DS

“R for Data Science” contains “data/heights.csv” in data import.

Is this reproducible?

Workarounds

data_url <- paste0("https://raw.githubusercontent.com/",
                   "rstudio/r2d3/master/vignettes/gallery/",
                   "bubbles/flare.csv")

# automatically download
download.file(data_url)

# automatically download only once
if (!file.exists("flare.csv"))
  download.file(data_url, "flare.csv")

Add to .gitignore? Upstream changes? Share across projects?

Caching with Pins

With pins we can easily cache resources,

pins::pin(data_url)

"/Users/javierluraschi/Library/Caches/pins/local/flare/flare.csv"

But wait, there is more…

Intro

Functionality

You can use the pins package to:

• Pin remote resources locally to work offline and cache results with ease, pin() stores resources in boards which you can then retrieve with pin_get().
• Discover new resources across different boards using pin_find().
• Share resources on GitHub, Kaggle or RStudio Connect by registering new boards with board_register().
• Resources can be anything from CSV, JSON, or image files to arbitrary R objects.

What can I pin?

Anything!

# pin a remote file
pins::pin(data_url)

# pin a local file
writeLines("Hello World!", "data.txt")
pins::pin("data.txt")

# pin a data frame
pins::pin(mtcars)

# pin a regression model
pins::pin(lm(mpg ~ ., mtcars))

Where can I store pins?

Anywhere! – That implements the ‘board’ interface.

library(pins)
board_register(board = "local", name = "local")

pin(data_url, board = "local")

What is a board?

A storage location, like your local file systems, GitHub, Kaggle or RStudio Connect.

Pin

pins::pin(iris, board = "rsconnect")

Authentication

library(pins)

# register from RStudio
board_register_rsconnect(server = "https://my-rsc.com", account = "my-user")

# register from RStudio Connect
board_register_rsconnect(server = "https://my-rsc.com", key = Sys.getenv("RSC_API"))

Discover

pin_find("iris", board = "rsconnect")

Share

# pin a remote file
pins::pin(data_url, board = "rsconnect")

# pin a local file
pins::pin("data.txt", board = "rsconnect")

# pin a data frame
pins::pin(mtcars, board = "rsconnect")

# pin a regression model
pins::pin(lm(mpg ~ ., mtcars), board = "rsconnect")

Resources

A pin can be anything,

pins::pin(lm(mpg ~ ., mtcars), board = "rsconnect")

Extensions

A pin can also be extended!

Use Cases

There are many interesting use cases available for pins, from caching remote resources to creating data pipelines.

See rstudio.github.io/pins/articles/use-cases.