define_lasso

Helper function to define a lasso penalty regularization object.
See <code>define_penalty</code> for more details.

Fits hierarchical regularized regression models
to incorporate potentially informative external data, Weaver and Lewinger (2019) <doi:10.21105/joss.01761>.
Utilizes coordinate descent to efficiently fit regularized regression models both with and without
external information with the most common penalties used in practice (i.e. ridge, lasso, elastic net).
Support for standard R matrices, sparse matrices and big.matrix objects.

Garrett Weaver

xrnet

Hierarchical Regularized Regression

Dixin Shen

Juan Pablo Lewinger

define_lasso function

<dl><dt>num_penalty</dt>
<dd>number of penalty values to fit in grid. Default is 20.</dd>
<dt>penalty_ratio</dt>
<dd>ratio between minimum and maximum penalty for x.
Default is 1e-04 if \(n &gt; p\) and 0.01 if \(n &lt;= p\).</dd>
<dt>user_penalty</dt>
<dd>user-defined vector of penalty values to use in penalty
path.</dd>
<dt>custom_multiplier</dt>
<dd>variable-specific penalty multipliers to apply to
overall penalty.
Default is 1 for all variables. 0 is no penalization.</dd></dl>

Arguments

Define lasso regularization object for predictor and external data — define_lasso

<dl>

<dt>num_penalty</dt>
<dd>number of penalty values to fit in grid. Default is 20.</dd>


<dt>penalty_ratio</dt>
<dd>ratio between minimum and maximum penalty for x.
Default is 1e-04 if \(n &gt; p\) and 0.01 if \(n &lt;= p\).</dd>


<dt>user_penalty</dt>
<dd>user-defined vector of penalty values to use in penalty
path.</dd>


<dt>custom_multiplier</dt>
<dd>variable-specific penalty multipliers to apply to
overall penalty.
Default is 1 for all variables. 0 is no penalization.</dd>

</dl>

define_lasso: Define lasso regularization object for predictor and external data

Description

Usage

Value

Arguments