expInstantiate

Instatiate the methods in the experiment for each one of the different parameter configurations.

Usage

expInstantiate(e, parameters = NULL, removeUnary = TRUE)

Arguments

e
The experiment object to be instantiated
parameters
A vector indicating the parameters to be instantiaded. If NULL or default all parameters would be considered.
removeUnary
Boolean value indicating if the unary parameters will be used in an instantiation or if the column can be erased.

Value

an experiment object

Description

When performing statistical tests or summarizing an experiment for a given output variable there can be different parameter configuration for each interaction of method and problem. Once applied the desired transformations this function can be used to remove unary parameters from the experiment or to instantiate the methods for each configuration.

Details

If any method is instantiated the cartesian product of the method and the selected parameters is performed and included in the resulting experiment as the methods variable. The name of the corresponding value will indicate the name of the former method and the value of each parameter instantiated.

Examples

# Create an experiment from the wekaExperiment experiment <- expCreate(wekaExperiment, name="test-exp", parameter="fold") # We would like to reduce the fold parameter by its mean value. It becomes an # unary parameter. experiment <- expReduce(experiment, "fold", mean) # Now we instantiate the experiment by the featureSelection parameter and # remove the unary fold parameter expInstantiate(experiment, removeUnary=TRUE)
#Experiment name: test-exp #method: J48,no, NaiveBayes,no, OneR,no, RandomForest,no, J48,yes, NaiveBayes,yes, OneR,yes, RandomForest,yes #problem: anneal, audiology, balance-scale, car, glass, horsecolic, hypothyroid, ionosphere, liver-disorders, lymph, primary-tumor, soybean, vehicle, vote, vowel #parameters: *) featureSelection [no,yes] (instantiated) #outputs: accuracy, trainingTime

Back to top

Built by staticdocs. Styled with bootstrap.