loss.biglasso

internal

Extend lasso and elastic-net model fitting for large data sets that
cannot be loaded into memory. Designed to be more memory- and
computation-efficient than existing lasso-fitting packages like 'glmnet' and
'ncvreg', thus allowing the user to analyze big data with limited RAM
<doi:10.32614/RJ-2021-001>.

Patrick Breheny

biglasso

Extending Lasso Model Fitting to Big Data

Yaohui Zeng

Chuyi Wang

Tabitha Peter

loss.biglasso function

<dl><dt>y</dt>
<dd>The observed response vector.</dd>
<dt>yhat</dt>
<dd>The predicted response vector.</dd>
<dt>family</dt>
<dd>Either "gaussian" or "binomial", depending on the response.</dd>
<dt>eval.metric</dt>
<dd>The evaluation metric for the cross-validated error and
for choosing optimal <code>lambda</code>:<ul>
<li><code>"default"</code> for linear regression is MSE (mean squared error) and for logistic regression is
misclassification error.</li>
<li><code>"MAPE"</code>, for linear regression only, is the Mean Absolute Percentage Error.</li>
<li>"auc", for logistic regression, is the area under the receiver operating
characteristic curve (ROC).</li>
</ul></dd>
<dt>grouped</dt>
<dd>Whether to calculate loss for the entire CV fold (<code>TRUE</code>), or for predictions
individually. Must be <code>TRUE</code> when <code>eval.metric</code> is 'auc'.</dd></dl>

Arguments

Internal biglasso functions — loss.biglasso

<dl>

<dt>y</dt>
<dd>The observed response vector.</dd>


<dt>yhat</dt>
<dd>The predicted response vector.</dd>


<dt>family</dt>
<dd>Either "gaussian" or "binomial", depending on the response.</dd>


<dt>eval.metric</dt>
<dd>The evaluation metric for the cross-validated error and
for choosing optimal <code>lambda</code>:<ul>
<li><code>"default"</code> for linear regression is MSE (mean squared error) and for logistic regression is
misclassification error.</li>
<li><code>"MAPE"</code>, for linear regression only, is the Mean Absolute Percentage Error.</li>
<li>"auc", for logistic regression, is the area under the receiver operating
characteristic curve (ROC).</li>
</ul></dd>


<dt>grouped</dt>
<dd>Whether to calculate loss for the entire CV fold (<code>TRUE</code>), or for predictions
individually. Must be <code>TRUE</code> when <code>eval.metric</code> is 'auc'.</dd>

</dl>

Internal biglasso functions

loss.biglasso: Internal biglasso functions

Description

Usage

Arguments

Details