Getting Started

Tidymodels skills are specialized Claude Code skills that guide you through machine learning and package development tasks using the Tidymodels ecosystem. Each skill contains step-by-step guidance for your specific task, best practices from the Tidymodels community, complete working examples, and reference documentation for deeper learning.

Installation

Install Claude Code

Claude Code is Anthropic’s official CLI for Claude. It brings Claude’s capabilities directly to your terminal, allowing you to work with AI assistance in your development environment. Installation is quick and works across macOS, Linux, and Windows.

curl -fsSL https://claude.ai/install.sh | sh

After installation, verify it’s working:

claude --version
irm https://claude.ai/install.ps1 | iex

After installation, verify it’s working:

claude --version

For detailed installation instructions, troubleshooting, and alternative methods, visit the official Claude Code documentation.

Authentication

On first use, Claude Code will prompt you to authenticate with your Anthropic account. For more information about authentication, see the Claude Code authentication guide.

Installing Skills

To use Tidymodels skills in Claude Code, first add the Tidymodels skills marketplace:

/plugin marketplace add tidymodels/skills

Then install the skill collection you need:

User Skills →

Install user skills for building predictive models:

/plugin install tidymodels-users@tidymodels-skills

Available skills:

  • Tabular Data Machine Learning: Build predictive models with proper validation practices
    • Data splitting and spending
    • Cross-validation and resampling
    • Feature engineering
    • Model tuning and selection
    • Performance evaluation

Developer Skills →

Install developer skills for creating Tidymodels extensions:

/plugin install tidymodels-developers@tidymodels-skills

Available skills:

Once installed, skills are available in your conversation and Claude Code is ready to assist with your task.

Using Tidymodels skills

Let’s walk through using the Tabular Data ML skill for a typical modeling workflow:

  1. Start Claude Code in your project directory:

    cd my-analysis-project
claude

  2. Install the user skills (one-time setup):

    /plugin marketplace add tidymodels/skills
/plugin install tidymodels-users@tidymodels-skills

  3. Describe your task:

    I have a CSV file with customer data and want to predict churn. The dataset has 15 features including demographics, usage patterns, and service history.

  4. Claude Code will guide you through:

    • Creating an appropriate train/test split
    • Setting up cross-validation for model tuning
    • Building and comparing multiple models
    • Evaluating performance with proper metrics
    • Only touching the test set when you explicitly approve

  5. Iterate and refine: Ask questions, request visualizations, tune hyperparameters, and explore feature engineering—all following Tidymodels best practices.

Learn More

Key Principles

When working with Tidymodels skills, Claude Code follows these principles:

  1. Test set protection: Development happens on training data. Test sets are only evaluated with explicit user permission.
  2. Proper validation: Uses cross-validation and resampling for honest performance estimates.
  3. Tidymodels conventions: Follows ecosystem patterns for consistency and reproducibility.
  4. Best practices: Implements approaches recommended by the Tidymodels team.

Tips for Success

  1. Be specific about your data: Describe your features, outcome variable, and any domain constraints
  2. Ask questions: If you’re unsure about an approach, ask Claude Code to explain the reasoning
  3. Review outputs: Always examine generated code and results before making decisions
  4. Iterate: Start with a simple baseline, then refine based on results
  5. Use visualizations: Request plots to understand model performance and data patterns

Getting Help