get_Model: construct a given regression model thanks to a training set

In iqspr: Inverse Molecular Design

Description

Construct a given regression model thanks to a training set.

Usage

get_Model(X = NULL, Y = NULL, model_name = "elasticnet", params = NA)

Arguments

X	is a matrix of features (e.g. fingerprints and/or physical descriptors) issued from a training set.
Y	is a one-dimensional vector, or a matrix, of observables (e.g. targeted chemical properties) issued from a training set.
model_name	is the model's name to be built ("elasticnet" by default).
params	is a list of fixed parameters applied to the model (see get_Model_params for a detailed view of the default paramaters per model). For the "svm" model, a.k.a Support Vector Machine, Gaussian RBF or Laplace kernel ("rbfdot" or "laplacedot" respectively) are supported (limitation from the sigest function used as optimizer on the sigma parameter).

Value

the desired regression model optimized.

Examples

## Not run: elasticNet_model <- get_Model(X = X[training,], Y = Y[training,],
model = "elasticnet", params = list("alpha" = 0.5))
## End(Not run)

iqspr documentation built on Aug. 1, 2017, 9:02 a.m.