cvpen-class

Class of object returned by a cross-validation performed through
the <code>crossval</code> method.

class

Fits classical sparse regression models with
efficient active set algorithms by solving quadratic problems as described by
Grandvalet, Chiquet and Ambroise (2017) <doi:10.48550/arXiv.1210.2077>. Also provides a few
methods for model selection purpose (cross-validation, stability selection).

Julien Chiquet

quadrupen

Sparsity by Worst-Case Quadratic Penalties

cvpen-class function

<dl>
<dt><code>lambda1</code>:</dt>
<dd>vector of \(\lambda_1\)
(\(\ell_1\) or \(\ell_\infty\) penalty levels)
for which each cross-validation has been performed.</dd><dt><code>lambda2</code>:</dt>
<dd>vector (or scalar) of \(\ell_2\)-penalty levels for
which each cross-validation has been performed.</dd><dt><code>lambda1.min</code>:</dt>
<dd>level of \(\lambda_1\) that minimizes the
error estimated by cross-validation.</dd><dt><code>lambda1.1se</code>:</dt>
<dd>largest level of \(\lambda_1\) such as
the cross-validated error is within 1 standard error of the
minimum.</dd><dt><code>lambda2.min</code>:</dt>
<dd>level of \(\lambda_2\) that minimizes the
error estimated by cross-validation.</dd><dt><code>cv.error</code>:</dt>
<dd>a data frame containing the mean
cross-validated error and its associated standard error for each
values of <code>lambda1</code> and <code>lambda2</code>.</dd><dt><code>folds</code>:</dt>
<dd>list of <code>K</code> vectors indicating the folds
used for cross-validation.</dd><dt><code>beta.min</code>:</dt>
<dd>the vector of parameters obtained by
fitting the problem on the full data set <code>x</code> and <code>y</code> with
<code>lambda1.min</code> and <code>lambda2.min</code> penalties.</dd><dt><code>beta.1se</code>:</dt>
<dd>the vector of parameters obtained by
fitting the problem on the full data set <code>x</code> and <code>y</code> with
<code>lambda1.1se</code> and <code>lambda2.min</code> penalties.</dd></dl>The specific <code>plot,cvpen-method</code> method is documented.

Slots

Class "cvpen" — cvpen-class

<dl>
<dt><code>lambda1</code>:</dt>
<dd>vector of \(\lambda_1\)
(\(\ell_1\) or \(\ell_\infty\) penalty levels)
for which each cross-validation has been performed.</dd>

<dt><code>lambda2</code>:</dt>
<dd>vector (or scalar) of \(\ell_2\)-penalty levels for
which each cross-validation has been performed.</dd>

<dt><code>lambda1.min</code>:</dt>
<dd>level of \(\lambda_1\) that minimizes the
error estimated by cross-validation.</dd>

<dt><code>lambda1.1se</code>:</dt>
<dd>largest level of \(\lambda_1\) such as
the cross-validated error is within 1 standard error of the
minimum.</dd>

<dt><code>lambda2.min</code>:</dt>
<dd>level of \(\lambda_2\) that minimizes the
error estimated by cross-validation.</dd>

<dt><code>cv.error</code>:</dt>
<dd>a data frame containing the mean
cross-validated error and its associated standard error for each
values of <code>lambda1</code> and <code>lambda2</code>.</dd>

<dt><code>folds</code>:</dt>
<dd>list of <code>K</code> vectors indicating the folds
used for cross-validation.</dd>

<dt><code>beta.min</code>:</dt>
<dd>the vector of parameters obtained by
fitting the problem on the full data set <code>x</code> and <code>y</code> with
<code>lambda1.min</code> and <code>lambda2.min</code> penalties.</dd>

<dt><code>beta.1se</code>:</dt>
<dd>the vector of parameters obtained by
fitting the problem on the full data set <code>x</code> and <code>y</code> with
<code>lambda1.1se</code> and <code>lambda2.min</code> penalties.</dd>



</dl>

The specific <code>plot,cvpen-method</code> method is documented.

cvpen-class: Class "cvpen"

Description

Arguments

Slots

See Also