MachineShop (version 1.1.0)

tune: Model Tuning and Selection

Description

Evaluate a model over a grid of tuning parameters or a list of specified models and select the best one according to resample estimation of predictive performance.

Usage

tune(x, ...)
# S3 method for formula
tune(x, data, models, grid = 3, fixed = NULL,
  control = CVControl, metrics = NULL, stat = mean,
  maximize = TRUE, ...)
# S3 method for ModelFrame
tune(x, models, grid = 3, fixed = NULL,
  control = CVControl, metrics = NULL, stat = mean,
  maximize = TRUE, ...)
# S3 method for recipe
tune(x, models, grid = 3, fixed = NULL,
  control = CVControl, metrics = NULL, stat = mean,
  maximize = TRUE, ...)

Arguments

x

defined relationship between model predictors and an outcome. May be a ModelFrame containing a formula, data, and optionally case weights; a formula; or a recipe.

...

arguments passed to the metrics functions.

data

data.frame containing observed predictors and outcomes.

models

MLModel function, function name, object or list of the aforementioned elements, such as that returned by expand.model.

grid

data.frame containing parameter values at which to evaluate a single model supplied to models, the number of parameter-specific values to generate automatically if the model has a pre-defined grid, or a call to Grid. Ignored in the case of a list of models.

fixed

list of fixed parameter values to combine with those in grid.

control

MLControl object, control function, or character string naming a control function defining the resampling method to be employed.

metrics

function, one or more function names, or list of named functions to include in the calculation of performance metrics. The default performance metrics are used unless otherwise specified. Model selection is based on the first specified metric.

stat

function to compute a summary statistic on resampled values of the metric for model selection.

maximize

logical indicating whether to select the model having the maximum or minimum value of the performance metric. Set automatically if a package metrics function is explicitly specified for the model selection.

Value

MLModelTune class object that inherits from MLModel.

See Also

ModelFrame, recipe, modelinfo, expand.model, Grid, MLControl, fit, plot, summary

Examples

# NOT RUN {
## Survival response example
library(MASS)

fo <- medv ~ .

# User-specified grid
(gbmtune1 <- tune(fo, data = Boston, model = GBMModel,
                  grid = expand.grid(n.trees = c(25, 50, 100),
                                     interaction.depth = 1:3,
                                     n.minobsinnode = c(5, 10)),
                  control = CVControl(folds = 10, repeats = 5)))

# Automatically generated grid
(gbmtune2 <- tune(fo, data = Boston, model = GBMModel, grid = 3,
                  control = CVControl(folds = 10, repeats = 5)))

# Randomly sampled grid points
(gbmtune3 <- tune(fo, data = Boston, model = GBMModel,
                  grid = Grid(length = 1000, random = 10),
                  control = CVControl(folds = 10, repeats = 5)))

summary(gbmtune3)
plot(gbmtune3, type = "line")

gbmfit <- fit(fo, data = Boston, model = gbmtune3)
varimp(gbmfit)
# }
# NOT RUN {
# }