dials (version 0.0.2)

dropout: Parameter objects related to parametric models.

Description

These are objects that can be used for modeling, especially in conjunction with the parsnip package.

Usage

dropout
epochs
activation
mixture
penalty
rbf_sigma
prod_degree
num_terms
num_comp
cost
scale_factor
margin
degree
deg_free
hidden_units
batch_size

Value

Each object is generated by either new_quant_param or new_qual_param.

Format

An object of class quant_param (inherits from param) of length 7.

Details

These objects are pre-made parameter sets that are useful when the model is based on some type of slope/intercept model.

• penalty: The total amount of regularization used. This is used by parsnip::linear_reg() and parsnip::logistic_reg() with glmnet models.

• mixture: the proportion of L1 regularization in the model. (parsnip::linear_reg() and parsnip::logistic_reg())

• dropout: the parameter dropout rate. (parsnip:::mlp())

• epochs: the number of iterations of training. (parsnip:::mlp())

• activation: the type of activation function between network layers. (parsnip:::mlp())

• hidden_units: the number of hidden units in a network layer. (parsnip:::mlp())

• batch_size: the mini-batch size for neural networks.

• rbf_sigma: the sigma parameters of a radial basis function.

• cost: a cost value for SVM models.

• scale_factor: the polynomial and hyperbolic tangent kernel scaling factor.

• margin: the SVM margin parameter (e.g. epsilon in the insensitive-loss function for regression).

• degree: the polynomial degree.

• prod_degree: the number of terms to combine into interactions. A value of 1 implies an additive model. Useful for MARS models.

• num_terms: a nonspecific parameter for the number of terms in a model. This can be used with models that include feature selection, such as MARS.

• num_comp: the number of components in a model (e.g. PCA or PLS components).

• deg_free: a parameter for the degrees of freedom.