R/hai-skewed-features.R

In healthyR.ai: The Machine Learning and AI Modeling Companion to 'healthyR'

#' Get Skewed Feature Columns
#'
#' @author Steven P. Sandeson II, MPH
#'
#' @description Takes in a data.frame/tibble and returns a vector of names of
#' the columns that are skewed.
#'
#' @details Takes in a data.frame/tibble and returns a vector of names of the skewed
#' columns. There are two other parameters. The first is the `.threshold` parameter
#' that is set to the level of skewness you want in order to consider the column
#' too skewed. The second is `.drop_keys`, these are columns you don't want to be
#' considered for whatever reason in the skewness calculation.
#'
#' @param .data The data.frame/tibble you are passing in.
#' @param .threshold A level of skewness that indicates where you feel a column
#' should be considered skewed.
#' @param .drop_keys A c() character vector of columns you do not want passed to
#' the function.
#'
#' @examples
#' hai_skewed_features(mtcars)
#' hai_skewed_features(mtcars, .drop_keys = c("mpg", "hp"))
#' hai_skewed_features(mtcars, .drop_keys = "hp")
#'
#' @return
#' A character vector of column names that are skewed.
#'
#' @export
#'

hai_skewed_features <- function(.data, .threshold = 0.6, .drop_keys = NULL) {

  # Tidyeval ----
  threshold <- as.numeric(.threshold)
  drop_keys <- .drop_keys

  # Checks ----
  if (!is.numeric(threshold)) {
    rlang::abort(
      message = "The '.threshold' parameter must be numeric.",
      use_cli_format = TRUE
    )
  }

  if (!is.data.frame(.data)) {
    rlang::abort(
      message = "The '.data' parameter must be a data.frame/tibble.",
      use_cli_format = TRUE
    )
  }

  if (!is.null(drop_keys) & !is.character(drop_keys)) {
    rlang::abort(
      message = "If provided, the '.drop_keys' parameter must be a 'character'. Using
      something like '.drop_keys = c('key_1','key_2',...).",
      use_cli_format = TRUE
    )
  }

  # Data ----
  data_tbl <- tibble::as_tibble(.data)

  # Transforms
  skewed_feature_names <- data_tbl %>%
    dplyr::select(tidyselect::vars_select_helpers$where(is.numeric)) %>%
    purrr::map_df(hai_skewness_vec) %>%
    tidyr::pivot_longer(cols = dplyr::everything()) %>%
    dplyr::filter(!name %in% drop_keys) %>%
    dplyr::mutate(name = as.factor(name)) %>%
    dplyr::rename(key = name) %>%
    dplyr::filter(value >= threshold) %>%
    dplyr::pull(key) %>%
    as.character()

  # Return ----
  return(skewed_feature_names)
}