RStudio Add-in for DataLinter

An RStudio add-in that brings the power of DataLinter directly into your RStudio session.

DataLinter performs automated sanity checks and linting on data frames and modelling code (23 built-in linters covering missing values, outliers, type consistency, modelling assumptions, and more). This add-in lets you lint selected code (and optional data) with a single click from the Addins menu.

Table of Contents

• Features
• Prerequisites
• Installation
• Quick Start
• Troubleshooting
• Contributing
• License
• Acknowledgements
• References

Features

• Lint any selected R code directly from the editor.
• Automatically detects data = my_variable arguments and sends the data frame to the linter.
• Falls back to a global LINTER_DATA variable when no data= argument is present.
• Real-time feedback inside RStudio (via the add-in UI powered by rstudioapi).
• Zero-configuration integration with the official DataLinter Docker server.

See the full DataLinter documentation for details on the 23 linters and configuration options.

Prerequisites

• R ≥ 4.0 (tested with imports: rstudioapi, httr, rjson)
• RStudio (Desktop or Server) with the Addins menu enabled
• Docker (recommended and documented deployment for DataLinter)
• Up-to-date pak or devtools + rstudioapi

Installation

1. DataLinter server (Docker)

docker pull ghcr.io/zgornel/datalinter-compiled:latest

2. RStudio-addin package

Note: Make sure you have up-to-date stable versions of devtools or pak and rstudioapi installed before installing the package.

# Recommended (fast & dependency-aware)
pak::pak("zgornel/Rstudio-Addin-DataLinter")

# Alternative
devtools::install_github("zgornel/Rstudio-Addin-DataLinter")

After installation the add-in appears automatically under Addins → DATALINTER.

Quick Start

1. Start the DataLinter server (see below
2. In RStudio, define sample data under the LINTER_DATA variable:

LINTER_DATA <- mtcars   # or any data.frame

3. Select any modelling code (or nothing) and choose Addins → DATALINTER / Lint data and code.
4. Review the lint report that appears.

Starting the DataLinter server

The add-in requires the DataLinter server running on port 10000 (default). The server can be started with

docker run -it --rm -p 10000:10000 \
  ghcr.io/zgornel/datalinter-compiled:latest \
    /datalinterserver/bin/datalinterserver \
      -i 0.0.0.0 \
      --config-path /datalinter/config/r_modelling_config.toml \
      --log-level info

If the server starts correctly, it should display something like:

[ Info: • Data linting server online @0.0.0.0:10000...
[ Info: Listening on: 0.0.0.0:10000, thread id: 1

How the add-in works

• Selected code is analyzed by the plugin
• If the code contains data = my_variable, that object is automatically serialized and included.
• Code and data are sent to the server
• If no code is selected, the plugin looks for a global LINTER_DATA variable.
• The plugin waits for an answer from the server and prints when received

Troubleshooting

• Server not reachable → Check Docker is running, port 10000 is free (docker ps), and firewall allows localhost traffic
• Port conflict → Change the published port (-p 10001:10000) and update the add-in configuration
• Large datasets → DataLinter is designed for typical modelling data; very large objects may need custom work
• Windows / macOS → Ensure Docker Desktop is running and “File Sharing” includes your project folders if using volumes -No output → Verify the server log shows successful connection; try increasing logging level with --log-level debug

Report issues via GitHub Issues.

Contributing

Please file an issue to report a bug or request a feature. See the parent DataLinter contributing guidelines as well.

License

This project is licensed under the GNU General Public License v3.0 (GPL-3) – see LICENSE for details. The underlying DataLinter engine uses an MIT license.

Acknowledgements

The initial version of DataLinter was fully inspired by this work written by Google brain research.

References

[1] https://en.wikipedia.org/wiki/Lint_(software)

[2] N. Hynes, D. Sculley, M. Terry "The data linter: Lightweight, automated sanity checking for ml data sets", NIPS MLSys Workshop, 2017; paper

[3] The data-linter code repository