easyml (version 0.1.0)

easy_support_vector_machine: Easily build and evaluate a support vector machine regression model.

Description

This function wraps the easyml core framework, allowing a user to easily run the easyml methodology for a support vector machine model.

Usage

easy_support_vector_machine(.data, dependent_variable, family = "gaussian",
  resample = NULL, preprocess = preprocess_scale, measure = NULL,
  exclude_variables = NULL, categorical_variables = NULL,
  train_size = 0.667, foldid = NULL, n_samples = 1000,
  n_divisions = 1000, n_iterations = 10, random_state = NULL,
  progress_bar = TRUE, n_core = 1, coefficients = FALSE,
  variable_importances = FALSE, predictions = TRUE,
  model_performance = TRUE, model_args = list())

Arguments

.data

A data.frame; the data to be analyzed.

dependent_variable

A character vector of length one; the dependent variable for this analysis.

family

A character vector of length one; the type of regression to run on the data. Choices are one of c("gaussian", "binomial"). Defaults to "gaussian".

resample

A function; the function for resampling the data. Defaults to NULL.

preprocess

A function; the function for preprocessing the data. Defaults to NULL.

measure

A function; the function for measuring the results. Defaults to NULL.

exclude_variables

A character vector; the variables from the data set to exclude. Defaults to NULL.

categorical_variables

A character vector; the variables that are categorical. Defaults to NULL.

train_size

A numeric vector of length one; specifies what proportion of the data should be used for the training data set. Defaults to 0.667.

foldid

A vector with length equal to length(y) which identifies cases belonging to the same fold.

n_samples

An integer vector of length one; specifies the number of times the coefficients and predictions should be generated. Defaults to 1000.

n_divisions

An integer vector of length one; specifies the number of times the data should be divided when replicating the measures of model performance. Defaults to 1000.

n_iterations

An integer vector of length one; during each division, specifies the number of times the predictions should be generated. Defaults to 10.

random_state

An integer vector of length one; specifies the seed to be used for the analysis. Defaults to NULL.

progress_bar

A logical vector of length one; specifies whether to display a progress bar during calculations. Defaults to TRUE.

n_core

An integer vector of length one; specifies the number of cores to use for this analysis. Currently only works on Mac OSx and Unix/Linux systems. Defaults to 1.

coefficients

A logical vector of length one; whether or not to generate coefficients for this analysis.

variable_importances

A logical vector of length one; whether or not to generate variable importances for this analysis.

predictions

A logical vector of length one; whether or not to generate predictions for this analysis.

model_performance

A logical vector of length one; whether or not to generate measures of model performance for this analysis.

model_args

A list; the arguments to be passed to the algorithm specified.

Value

A list of class easy_support_vector_machine.

See Also

Other recipes: easy_analysis, easy_avNNet, easy_deep_neural_network, easy_glinternet, easy_glmnet, easy_neural_network, easy_random_forest

Examples

library(easyml) # https://github.com/CCS-Lab/easyml

# Gaussian
data("prostate", package = "easyml")
results <- easy_support_vector_machine(prostate, "lpsa", 
                                       n_samples = 10, 
                                       n_divisions = 10, 
                                       n_iterations = 2, 
                                       random_state = 1, n_core = 1)

# Binomial
data("cocaine_dependence", package = "easyml")
results <- easy_support_vector_machine(cocaine_dependence, "diagnosis", 
                                       family = "binomial", 
                                       preprocesss = preprocess_scale, 
                                       exclude_variables = c("subject"), 
                                       categorical_variables = c("male"), 
                                       n_samples = 10, 
                                       n_divisions = 10, 
                                       n_iterations = 2, 
                                       random_state = 1, n_core = 1)