Implement EDM Model Families, R User Workflows, and Vignette Documentation

[aman-raj-srivastva]

Goal

Build upon the delay embedding infrastructure #1528, prediction functors #1561, and FIMS/TMB integration #1620, this final phase delivers high-level R-facing interfaces, model family abstractions (univariate and multivariate), expanded R test coverage, and comprehensive user documentation for Empirical Dynamic Modeling (EDM) within FIMS.

Tasks

• Implement high-level R wrapper functions (e.g., fit_edm(), edm_forecast()) to simplify running EDM models without directly interacting with low-level C++ interfaces.
• Implement standalone univariate and multivariate EDM model family abstractions
• Demonstrate integration of EDM predictions (e.g., empirical recruitment and biomass forecasting) with traditional FIMS stock assessment workflows.
• Expand testthat coverage for:
  • High-level R wrapper functions
  • FIMSFrame delay embedding extraction utilities
  • EDM model output reshaping and validation
• Create a user-facing vignette (vignettes/edm-user-guide.Rmd) demonstrating:
  • Data preparation using FIMSFrame
  • Model configuration
  • Model fitting
  • Forecast generation
  • Visualization of results
• Add complete roxygen2 documentation for exported R functions.
• Add Doxygen documentation for all new C++ classes and methods.

Notes

This issue represents the final implementation phase of the GSoC 2026 EDM project.