splendid_process: Process data

In AlineTalhouk/splendid: SuPervised Learning ENsemble for Diagnostic IDentification

Process data

Description

Process the data by converting categorical predictors to dummy variables, standardizing continuous predictors, and apply subsampling techniques.

Usage

splendid_process(
  data,
  class,
  algorithms,
  convert = FALSE,
  standardize = FALSE,
  sampling = c("none", "up", "down", "smote"),
  seed_samp = NULL
)

Arguments

data	data frame with rows as samples, columns as features
class	true/reference class vector used for supervised learning
algorithms	character vector of algorithms to use for supervised learning. See Algorithms section for possible options. By default, this argument is NULL, in which case all algorithms are used.
convert	logical; if TRUE, converts all categorical variables in data to dummy variables. Certain algorithms only work with such limitations (e.g. LDA).
standardize	logical; if TRUE, the training sets are standardized on features to have mean zero and unit variance. The test sets are standardized using the vectors of centers and standard deviations used in corresponding training sets.
sampling	the default is "none", in which no subsampling is performed. Other options include "up" (Up-sampling the minority class), "down" (Down-sampling the majority class), and "smote" (synthetic points for the minority class and down-sampling the majority class). Subsampling is only applicable to the training set.
seed_samp	random seed used for reproducibility in subsampling training sets for model generation

Details

If all the variables in the original data are already continuous, nothing is done. Otherwise, conversion is performed if convert = TRUE using dummify(). An error message is thrown if there are categorical variables and convert = FALSE, indicating exactly which algorithms specified require data conversion. Classification algorithms LDA and the MLR family have such a limitation.

Continuous predictors can be scaled to have zero mean and unit variance with standardize = TRUE. Dummy variables coded to 0 or 1 are never standardized.

Subsampling techniques can be applied with sampling methods passed to subsample().

Value

A pre-processed data frame for model training

Author(s)

Derek Chiu

See Also

dummify(), subsample()

Examples

data(hgsc)
cl <- attr(hgsc, "class.true")

# Nothing happens if data is all continuous
data_same <- splendid_process(hgsc, class = cl, algorithms = "lda", convert =
TRUE)
identical(hgsc, data_same)

# Dummy variables created if there are categorical variables
data_dummy <- splendid_process(iris, class = iris$Species, algorithms =
"lda", convert = TRUE)
head(data_dummy)

# Some algorithms are robust to the covariate data structure
data_robust <- splendid_process(iris, class = iris$Species, algorithms =
"rf", convert = FALSE)
identical(iris, data_robust)

# Standardize and down-sample
iris2 <- iris[1:130, ]
data_scale_down <- splendid_process(iris2, class = iris2$Species, algorithms
= "rf", standardize = TRUE, sampling = "down")
dim(data_scale_down)

# Other algorithms require conversion
## Not run: 
splendid_process(iris, class = iris$Species, algorithms = "lda", convert =
FALSE)

## End(Not run)

AlineTalhouk/splendid documentation built on Feb. 23, 2024, 9:37 p.m.