R Data Analysis Workflow

Comprehensive data analysis workflow using tidy principles. This command orchestrates multiple agents to perform end-to-end data analysis.

Workflow Steps

1. Data Assessment

First, the data-wrangler agent will:

• Load and inspect the data structure
• Identify variable types and distributions
• Check for missing values and data quality issues
• Document data cleaning requirements

2. Data Preparation

The feature-engineer agent will:

• Create preprocessing recipe based on data characteristics
• Handle missing data appropriately
• Encode categorical variables
• Create derived features if needed
• Document all transformations

3. Exploratory Analysis

The viz-specialist agent will:

• Create distribution plots for key variables
• Generate correlation analyses
• Produce visualizations for relationships of interest
• Create publication-ready figures

4. Statistical Analysis

Based on the analysis type, the appropriate agent will be engaged:

• biostatistician for inferential statistics, survival analysis, clinical endpoints
• tidymodels-engineer for predictive modeling and machine learning

5. Results Documentation

The reporting-engineer agent will:

• Compile analysis results into a reproducible report
• Include all code, outputs, and interpretations
• Generate appropriate output format (HTML, PDF, Word)

Usage

/r-analysis [data_path] [analysis_type] [output_format]

Parameters

• data_path: Path to the dataset (CSV, RDS, Excel, etc.)
• analysis_type: One of descriptive, inferential, predictive, survival, clinical
• output_format: One of html, pdf, word (default: html)

Output Location

All outputs will be written to:

output/
├── code/           # R scripts for reproducibility
├── reports/        # Analysis report
├── figures/        # Generated visualizations
└── models/         # Saved model objects (if applicable)

Example

/r-analysis data/patient_outcomes.csv survival html

This will perform survival analysis on the patient outcomes data and generate an HTML report.

Notes

• All code is generated in the output folder - your original data files are never modified
• The workflow adapts based on data characteristics discovered during assessment
• Interactive mode will ask clarifying questions about analysis goals