Flexible Rank-Preserving Correlation Engine

Implements a fast, flexible method for simulating continuous variables with specified rank correlations using the Iman–Conover transformation (Iman & Conover, 1982 ) and back-ranking. Includes plotting tools and error-diagnostics.

flexIC

flexIC is a high-precision Iman–Conover engine for generating continuous variables that preserve rank correlation with marginal fidelity. It offers tunable convergence control, allowing you to aggressively reduce rank-correlation distortion—at the cost of a few extra milliseconds.

Use it to:

• Simulate data with a target Spearman or Kendall structure
• Preserve original variable distributions via back-ranking
• Validate or stress-test statistical methods under structured dependence

🚀 Why use flexIC?

Most Iman–Conover implementations:

• Run once with no convergence check
• Do not guarantee low error
• Break marginal shapes in edge cases

flexIC:

• Iterates until max abs rank-correlation error ≤ ε
• Keeps original marginal shapes intact
• Returns detailed error diagnostics
• Finishes in milliseconds on typical datasets

📦 Installation

# Development version (until on CRAN)
remotes::install_github("TheotherDrWells/flexIC")