Modeltime H2O Roadmap

[mdancho84]

A working roadmap for developing an H2O extension for modeltime.

MVP Package:

• Create Package Structure
• Develop MVP Algorithms (Develop H2O Regression Algorithms #2) - Get basic use-cases from Shafi, Develop MVP algorithms, Test on several datasets
• Develop R
  • Dependencies (DESCRIPTION - Add Package Dependencies #3 )
  • Core automl_reg() algorithm (Develop Core AutoML Algorithm #4)
  • Saving/Loading Models / Workflows (Saving / Loading H2O Models #5)
• Create Tests (Create Tests #7)
• Automate Tests with GitHub Actions (Add GitHub Actions #8)
• Make AWESOME README (Make an Awesome README #9)
• Add Vignettes - Getting Started (Getting Started Vignette #10)
• Make Hex Logo (Make Hex Logo #11)
• Add Pkgdown Documentation (Create Pkgdown Website #12)
• Finalize DESCRIPTION
• Send to CRAN

Future efforts

• More Efficient Refitting - Use parameters from specific model rather than fully re-training AutoML (More Efficient Refitting: modeltime_refit() retrains all models. Allow user to retrain only the best model.  #14)
• H2O Leaderboard - Allow users to view the Leaderboard and swap out models (H2O Leaderboard Control: Add functions to select and swap preferred models #15)