Learn R Programming

NumericEnsembles (version 1.0.0)

Numeric: Numeric—function to automatically build 18 individual models and 14 ensembles then return the results to the user

Description

Numeric—function to automatically build 18 individual models and 14 ensembles then return the results to the user

Usage

Numeric(
  data,
  colnum,
  numresamples,
  remove_VIF_above = 5,
  remove_data_correlations_greater_than = 0.99,
  remove_ensemble_correlations_greater_than = 0.98,
  scale_all_predictors_in_data = c("Y", "N"),
  data_reduction_method = c(0("none"), 1("BIC exhaustive"), 2("BIC forward"),
    3("BIC backward"), 4("BIC seqrep"), 5("Mallows_cp exhaustive"),
    6("Mallows_cp forward"), 7("Mallows_cp backward"), 8("Mallows_cp, seqrep")),
  ensemble_reduction_method = c(0("none"), 1("BIC exhaustive"), 2("BIC forward"),
    3("BIC backward"), 4("BIC seqrep"), 5("Mallows_cp exhaustive"),
    6("Mallows_cp forward"), 7("Mallows_cp backward"), 8("Mallows_cp, seqrep")),
  how_to_handle_strings = c(0("none"), 1("factor levels"), 2("One-hot encoding"),
    3("One-hot encoding with jitter")),
  predict_on_new_data = c("Y", "N"),
  set_seed = c("Y", "N"),
  save_all_trained_models = c("Y", "N"),
  save_all_plots = c("Y", "N"),
  use_parallel = c("Y", "N"),
  stratified_random_column,
  train_amount,
  test_amount,
  validation_amount
)

Value

a real number

Arguments

data

data can be a CSV file or within an R package, such as MASS::Boston

colnum

a column number in your data

numresamples

the number of resamples

remove_VIF_above

remove columns with Variable Inflation Factor above value chosen by the user

remove_data_correlations_greater_than

maximum value for correlations of the original data (such as the Boston Housing data set)

remove_ensemble_correlations_greater_than

maximum value for correlations of the ensemble

scale_all_predictors_in_data

"Y" or "N" to scale numeric data

data_reduction_method

0(none), BIC (1, 2, 3, 4) or Mallow's_cp (5, 6, 7, 8) for Forward, Backward, Exhaustive and SeqRep

ensemble_reduction_method

0(none), BIC (1, 2, 3, 4) or Mallow's_cp (5, 6, 7, 8) for Forward, Backward, Exhaustive and SeqRep

how_to_handle_strings

0: No strings, 1: Factor values, 2: One-hot encoding, 3: One-hot encoding AND jitter

predict_on_new_data

"Y" or "N". If "Y", then you will be asked for the new data

set_seed

"Y" or "N" to set the seed to make the results fully reproducible

save_all_trained_models

"Y" or "N". If "Y", then places all the trained models in the temporary directory, tempdir(), and the trained models may be retreived from there

save_all_plots

Saves all plots to the tempdir() directory, and the plots may be retreived from there

use_parallel

"Y" or "N" for parallel processing

stratified_random_column

0 if no stratified random sampling, or column number for stratified random sampling

train_amount

set the amount for the training data

test_amount

set the amount for the testing data

validation_amount

Set the amount for the validation data