R/ml_feature_feature_hasher.R

Defines functions validator_ml_feature_hasher new_ml_feature_hasher ft_feature_hasher.tbl_spark ft_feature_hasher.ml_pipeline ft_feature_hasher.spark_connection ft_feature_hasher

Documented in ft_feature_hasher

#' Feature Transformation -- FeatureHasher (Transformer)
#'
#' @details Feature hashing projects a set of categorical or numerical features into a
#'   feature vector of specified dimension (typically substantially smaller than
#'   that of the original feature space). This is done using the hashing trick
#'   \url{https://en.wikipedia.org/wiki/Feature_hashing} to map features to indices
#'   in the feature vector.
#'
#'   The FeatureHasher transformer operates on multiple columns. Each column may
#'     contain either numeric or categorical features. Behavior and handling of
#'     column data types is as follows: -Numeric columns: For numeric features,
#'     the hash value of the column name is used to map the feature value to its
#'     index in the feature vector. By default, numeric features are not treated
#'     as categorical (even when they are integers). To treat them as categorical,
#'     specify the relevant columns in categoricalCols. -String columns: For
#'      categorical features, the hash value of the string "column_name=value"
#'      is used to map to the vector index, with an indicator value of 1.0.
#'      Thus, categorical features are "one-hot" encoded (similarly to using
#'      OneHotEncoder with drop_last=FALSE). -Boolean columns: Boolean values
#'      are treated in the same way as string columns. That is, boolean features
#'      are represented as "column_name=true" or "column_name=false", with an
#'      indicator value of 1.0.
#'
#'  Null (missing) values are ignored (implicitly zero in the resulting feature vector).
#'
#'  The hash function used here is also the MurmurHash 3 used in HashingTF. Since a
#'  simple modulo on the hashed value is used to determine the vector index, it is
#'  advisable to use a power of two as the num_features parameter; otherwise the
#'  features will not be mapped evenly to the vector indices.
#'
#' @param input_cols Names of input columns.
#' @param output_col Name of output column.
#' @param num_features Number of features. Defaults to \eqn{2^18}.
#' @param categorical_cols Numeric columns to treat as categorical features.
#'   By default only string and boolean columns are treated as categorical,
#'   so this param can be used to explicitly specify the numerical columns to
#'   treat as categorical.
#' @template roxlate-ml-feature-transformer
#'
#' @export
ft_feature_hasher <- function(x, input_cols = NULL, output_col = NULL,
                              num_features = 2^18, categorical_cols = NULL,
                              uid = random_string("feature_hasher_"), ...) {
  check_dots_used()
  UseMethod("ft_feature_hasher")
}

ml_feature_hasher <- ft_feature_hasher

#' @export
ft_feature_hasher.spark_connection <- function(x, input_cols = NULL, output_col = NULL,
                                               num_features = 2^18, categorical_cols = NULL,
                                               uid = random_string("feature_hasher_"), ...) {
  .args <- list(
    input_cols = input_cols,
    output_col = output_col,
    num_features = num_features,
    categorical_cols = categorical_cols,
    uid = uid
  ) %>%
    c(rlang::dots_list(...)) %>%
    validator_ml_feature_hasher()

  jobj <- spark_pipeline_stage(
    x, "org.apache.spark.ml.feature.FeatureHasher",
    input_cols = .args[["input_cols"]], output_col = .args[["output_col"]], uid = .args[["uid"]]
  ) %>%
    invoke("setNumFeatures", .args[["num_features"]]) %>%
    jobj_set_param("setCategoricalCols", .args[["categorical_cols"]])

  new_ml_feature_hasher(jobj)
}

#' @export
ft_feature_hasher.ml_pipeline <- function(x, input_cols = NULL, output_col = NULL,
                                          num_features = 2^18, categorical_cols = NULL,
                                          uid = random_string("feature_hasher_"), ...) {
  stage <- ft_feature_hasher.spark_connection(
    x = spark_connection(x),
    input_cols = input_cols,
    output_col = output_col,
    num_features = num_features,
    categorical_cols = categorical_cols,
    uid = uid,
    ...
  )
  ml_add_stage(x, stage)
}

#' @export
ft_feature_hasher.tbl_spark <- function(x, input_cols = NULL, output_col = NULL,
                                        num_features = 2^18, categorical_cols = NULL,
                                        uid = random_string("feature_hasher_"), ...) {
  stage <- ft_feature_hasher.spark_connection(
    x = spark_connection(x),
    input_cols = input_cols,
    output_col = output_col,
    num_features = num_features,
    categorical_cols = categorical_cols,
    uid = uid,
    ...
  )
  ml_transform(stage, x)
}

new_ml_feature_hasher <- function(jobj) {
  new_ml_transformer(jobj, class = "ml_feature_hasher")
}

validator_ml_feature_hasher <- function(.args) {
  .args[["input_cols"]] <- cast_nullable_string_list(.args[["input_cols"]])
  .args[["output_col"]] <- cast_nullable_string(.args[["output_col"]])
  .args[["categorical_cols"]] <- cast_nullable_string_list(.args[["categorical_cols"]])
  .args[["num_features"]] <- cast_scalar_integer(.args[["num_features"]])
  .args[["uid"]] <- cast_scalar_character(.args[["uid"]])
  .args
}

Try the sparklyr package in your browser

Any scripts or data that you put into this service are public.

sparklyr documentation built on Jan. 8, 2022, 5:06 p.m.