tune: Tune parameters of modeling procedures

In emil: Evaluation of Modeling without Information Leakage

Description

These functions are rarely needed to be called manually as they are automatically called by fit and evaluate when needed.

Usage

tune(procedure, ..., .verbose = getOption("emil_verbose", FALSE))

is_tuned(procedure)

is_tunable(procedure)

detune(procedure)

Arguments

procedure	Modeling procedure, or list of modeling procedures, as produced by modeling_procedure.
...	Sent to evaluate.
.verbose	Whether to print an activity log. Set to -1 to suppress all messages.

Value

A tuned modeling procedures or a list of such.

Logical indicating if the procedure(s) are tuned.

Logical indicating if the has tunable parameters.

A list of untuned modeling procedures.

Author(s)

Christofer Bäcklin

See Also

emil, modeling_procedure, evaluate, fit, predict, get_importance

Examples

procedure <- modeling_procedure("randomForest", parameter=list(mtry=1:4))
tuned.procedure <- tune(procedure, x=iris[-5], y=iris$Species)
mod <- fit(tuned.procedure, x=iris[-5], y=iris$Species)

Example output

Attaching package: 'emil'

The following object is masked from 'package:base':

    mode

Loading required namespace: randomForest

emil documentation built on Aug. 1, 2018, 1:03 a.m.