README.Rmd

---
output: github_document
---

<!-- README.md is generated from README.Rmd. Please edit that file -->

```{r setup, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)
```

# witchcraft <a href='https://bigdata.cs.ut.ee/'><img src='man/figures/logo.png' align="right" height="138.5" /></a>
<!-- badges: start -->
[![CRAN_Status_Badge](http://www.r-pkg.org/badges/version/witchcraft)](https://cran.r-project.org/package=witchcraft)
[![lifecycle](https://img.shields.io/badge/lifecycle-experimental-orange.svg)](https://www.tidyverse.org/lifecycle/#experimental)
  [![Travis build status](https://travis-ci.org/brurucy/witchcraft.svg?branch=master)](https://travis-ci.org/brurucy/witchcraft)
<!-- badges: end -->

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](https://arxiv.org/pdf/1603.06560.pdf) 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](https://cran.r-project.org/package=witchcraft) will be possible:

```{r cran-installation, eval = FALSE}
install.packages("witchcraft")
```

You can always install the **development** version from
[GitHub](https://github.com/brurucy/witchcraft)

```{r gh-installation, eval = FALSE}
# install.packages("remotes")
remotes::install_github("brurucy/witchcraft")
```

Installing this software requires a compiler.

## Valid example

```{r example, message=FALSE, eval=FALSE}
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

```