{"Package":"featuretoolsR","Type":"Package","Title":"Interact with the 'Python' Module 'Featuretools'","Version":"0.4.4","Authors@R":"person(\"Magnus\", \"Furugrd\", email = \"magnus.furugard@gmail.com\", role = c(\"aut\", \"cre\"))","Maintainer":"Magnus Furugrd ","Description":"A 'reticulate'-based interface to the 'Python' module 'Featuretools'.\nThe package grants functionality to interact with 'Pythons' 'Featuretools' module, which allows\nfor automated feature engineering on any data frame. Valid features and new data sets can, after\nfeature synthesis, easily be extracted.","License":"MIT + file LICENSE","URL":"https://github.com/magnusfurugard/featuretoolsR","BugReports":"https://github.com/magnusfurugard/featuretoolsR/issues","Depends":"R (>= 3.4.2)","Imports":"reticulate, caret, dplyr, purrr, stringr, tibble, magrittr,\ncli, testthat, rstudioapi","Encoding":"UTF-8","LazyData":"true","RoxygenNote":"6.1.1","NeedsCompilation":"no","Packaged":"2020-04-25 09:43:57 UTC; Magnus","Author":"Magnus Furugrd [aut, cre]","Repository":"CRAN","Date/Publication":"2020-04-25 16:50:02 UTC","repoType":"cran","tarballUrl":"ftp://cran.r-project.org/pub/R/src/contrib/featuretoolsR_0.4.4.tar.gz","jsonAuthors":[{"name":"Magnus Furugrd","email":"magnus.furugard@gmail.com","maintainer":true}],"readme":"# featuretoolsR\nAn R interface to the Python module Featuretools.\n\n# General\n`featuretoolsR` provides functionality from the Python module `featuretools`, which aims to automate feature engineering. This package is very much a work in progress as Featuretools offers a lot of functionality. Any PRs are much appreciated.\n\n# Installing\n\n## Package\n### CRAN\nThe latest stable release is found on [CRAN](https://cran.r-project.org/package=featuretoolsR).\n\n### Github\nYou can get the latest version of `featuretoolsR` by installing it straight from Github: `devtools::install_github(\"magnusfurugard/featuretoolsR\")`.\n\n## Featuretools\nYou'll need to have a working Python environment as well as `featuretools` installed. The recommended way is to use the built-in function `install_featuretools()` which automatically sets up a virtual environment for the package and installs `featuretools`.\n\n# Usage\nAll functions in `featuretoolsR` comes with documentation, but it's advised to briefly browse through the [Featuretools Python documentation](https://docs.featuretools.com/). It'll cover things like `entities`, `relationships` and `dfs`. \n\n## Creating an entityset\nAn entityset is the set which contain all your entities. To create a set and add an entity straight away, you can use `as_entityset`. \n```\n# Libs\nlibrary(featuretoolsR)\nlibrary(magrittr)\n\n# Create some mock data\nset_1 <- data.frame(key = 1:100, value = sample(letters, 100, T), a = rep(Sys.Date(), 100))\nset_2 <- data.frame(key = 1:100, value = sample(LETTERS, 100, T), b = rep(Sys.time(), 100))\n\n# Create entityset\nes <- as_entityset(\n set_1, \n index = \"key\", \n entity_id = \"set_1\", \n id = \"demo\", \n time_index = \"a\"\n)\n```\n\n## Adding entities\nTo add entities (i.e if you have relational data across multiple `data.frames`), this can be achieved with `add_entity`. This function is pipe friendly. For this demo-case, we'll use `set_2`.\n```\nes <- es %>%\n add_entity(\n df = set_2, \n entity_id = \"set_2\", \n index = \"key\", \n time_index = \"b\"\n )\n```\n\n## Defining relationships\nWith relational data, it's useful to define a relationship between two or more entities. This can be done with `add_relationship`.\n```\nes <- es %>%\n add_relationship(\n parent_set = \"set_1\", \n child_set = \"set_2\", \n parent_idx = \"key\", \n child_idx = \"key\"\n )\n```\n\n## Deep feature synthesis\nThe bread and butter of Featuretools is the `dfs`-function (official docs [here](https://docs.featuretools.com/en/stable/automated_feature_engineering/afe.html)). It will attempt to create features based on `*_primitives` you provide (more on primitives below).\n```\nft_matrix <- es %>%\n dfs(\n target_entity = \"set_1\", \n trans_primitives = c(\"and\", \"cum_sum\")\n )\n```\n\n## Tidying up\nTo use the new data.frame/features created by `dfs`, a function unique for `featuretoolsR`, `tidy_feature_matrix` can be used. A few \"nice-to-have\" arguments can be passed to clean the new data, like removing near zero variance variables, as well as replacing `NaN` with `NA`.\n```\ntidy <- tidy_feature_matrix(ft_matrix, remove_nzv = T, nan_is_na = T, clean_names = T)\n```\n\n# Primitives\nFeaturetools supports a lot of primitives. These are accessible with the function `list_primitives()` which returns a data.frame containing type (aggregation (`agg_primitives`) or transform (`trans_primitives`)), name (in the example above, \"and\" and \"divide\") as well as a brief description of the primitive itself.\n\n# Credits\n[reticulate](https://github.com/rstudio/reticulate) - an R interface to Python.\n\n[Featuretools](https://github.com/Featuretools/featuretools)","jobInfo":{"package":"featuretoolsR","version":"0.4.4","parsingStatus":"success","parserVersion":1,"parsedAt":"2020-04-25T19:30:11+0000"}}

Description

Usage

Arguments

Value

Examples