GetModelComparisons: A function to build and return a ModelComparison object.

In orionw/BestModel: A package to run and compare standard machine learning models with easy visualization

Description

This function does not take any pre-built models, instead it creates them based on input to model.list.

Usage

GetModelComparisons(training.set, training.classes.input,
  validation = "80/20", model.list = "fast", trctrl = "none")

Arguments

training.set	The training data to build the models.
training.classes.input	The labels for the training data.
validation	The type of model validation: cross-validation ("cv"), or a training split (in the form XX/XX where XX is a two digit percent).
model.list	A vector or list of characters that describe what models to build. Models include "neuralnet", "svmlinear", "svmradial", "knn", "randomforest", 'glmnet", and 'glm." Keywords include "fast" (glmnet, svmlinear), "all", and "expensive" (neuralnet, svmradial, glmnet, and randomforest)
trctrl	The trctrl used for the caret package to train the models. Defaults to a basic version. Used for customized training options including parallelization, etc.

Value

A fully created ModelComparison object.

Examples

titanic <- PrepareNumericTitanic()

# create the models
comp <- GetModelComparisons(titanic[, -1], titanic[, 1], validation="80/20",
                            model.list="fast")

orionw/BestModel documentation built on Aug. 17, 2019, 7:29 p.m.