In DataSystemsGroupUT/SmartML: Machine Learning Automation

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

witchcraft

The R package witchcraft is an opinionated framework for automated machine learning, with the intent of being frequently updated with the newest state-of-the-art optimization methods.

At the moment, witchcraft uses the Bayesian-Optimization-Hyperband algorithm.

Besides Combined Algorithm Selection and Hyperparameter optimization, witchcraft provides tools to evaluate the results, which are consistent with the mlr3 workflow.

Installation

Soon, installing the stable version from CRAN will be possible:

install.packages("witchcraft")

You can always install the development version from GitHub

# install.packages("remotes")
remotes::install_github("brurucy/witchcraft")

Installing this software requires a compiler.

Valid example

library(SmartML)
library(readr)

data_train <- readr::read_csv('inst/extdata/dota_train.csv') %>%
  as.data.table()

data_test <- readr::read_csv('inst/extdata/dota_test.csv') %>%
  as.data.table()

data_train[, class := factor(class, levels = unique(class)) %>% sort()]
data_test[, class := factor(class, levels = unique(class)) %>% sort()]

params <- SmartML:::get_random_hp_config('kknn', columns = ncol(data_train) - 1)

print(typeof(params$kernel))
params