/rtemis

Advanced Machine Learning and Visualization

Primary LanguageR

rtemis Machine Learning and Visualization Build Status

A platform for advanced Machine Learning research and applications.
The goal of rtemis is to make data science efficient and accessible with no compromise on flexibility.

Documentation

Requirements

R version 4.1 or higher

Installation

See here for more setup and installation instructions.

install.packages("remotes")
remotes::install_github("egenn/rtemis")

Note: Make sure to keep your installation updated by running remotes::install_github("egenn/rtemis") regularly: it will only proceed if there are updates available

60-second intro to rtemis

Install dependencies if they are not already installed:

packages <- c("pbapply", "ranger")
.add <- !packages %in% installed.packages()
install.packages(packages[.add])

Load rtemis and get cross-validated random forest performance on the iris dataset:

library(rtemis)
mod <- elevate(iris)
mod$plot()

What's new

0.82

  • Themes: New darkgray theme now always on whether you like it or not - jk: it's the new default but you can always set your own default using e.g. options(rt.theme = "lightgrid"). Also, new lightgray theme.
  • New option to set default plotting font: e.g. options(rt.font = "Fira Sans")
  • Many improvements / additions to dplot3* functions.
  • Plenty more I haven't had a chance to document here

0.80.0

An accumulation of updates and added functionality, algorithms, graphics.
Majority of mplot3 and dplot3 functions now work with the new theme system provided by theme_* functions like theme_lightgrid and theme_darkgrid.

0.79

07-02-2019: "Super Papaya" Release out

0.78

04-02-2019: rtemis moved to public repo

Features

  • Visualization

    • Static: mplot3 family (base graphics)
    • Dynamic: dplot3 family (plotly)

  • Unsupervised Learning

    • Clustering: u.*
    • Decomposition: d.*

  • Supervised Learning

    • Classification, Regression, Survival Analysis: s.*

  • Cross-Decomposition

    • Sparse Canonical Correlation / Sparse Decomposition: x.*

  • Meta-Models

    • Model Stacking: metaMod()
    • Modality Stacking: metaFeat()
    • Group-weighted Stacking: metaGroup()

    (metaFeat and metaGroup have been removed for updating)

Ongoing work

  • rtemis is under active development
  • Novel algorithms developed in rtemis will generally be added to this public repository around the publication of the corresponding papers.
  • R Documentation is ongoing.



2021 Efstathios (Stathis) D. Gennatas MBBS AICSM PhD