get.ideal.penalty

This function takes a fitted <code>linear.logic</code> model and independent
validation data as input for finding the ideal LASSO complexity penalty
<code>s</code>.

A statistical learning method that tries to find the best set
of predictors and interactions between predictors for modeling binary or
quantitative response data in a decision tree. Several search algorithms
and ensembling techniques are implemented allowing for finetuning the
method to the specific problem. Interactions with quantitative
covariables can be properly taken into account by fitting local
regression models. Moreover, a variable importance measure for assessing
marginal and interaction effects is provided. Implements the
procedures proposed by Lau et al. (2024, <doi:10.1007/s10994-023-06488-6>).

Michael Lau

logicDT

Identifying Interactions Between Binary Predictors

get.ideal.penalty function

<dl><dt>model</dt>
<dd>A fitted <code>linear.logic</code> model (i.e., a model created via
<code>fitLinearLogicModel</code> or <code>fitLinearBoostingModel</code>)</dd>
<dt>X</dt>
<dd>Matrix or data frame of binary input data.
This object should correspond to the binary matrix for fitting the model.</dd>
<dt>y</dt>
<dd>Response vector. 0-1 coding for binary outcomes.</dd>
<dt>Z</dt>
<dd>Optional quantitative covariables supplied as a matrix or
data frame. Only used (and required) if the model was fitted using them.</dd>
<dt>scoring_rule</dt>
<dd>The scoring rule for evaluating the validation
error and its standard error. For classification tasks, <code>"deviance"</code>
or <code>"Brier"</code> should be used.</dd>
<dt>choose</dt>
<dd>Model selection scheme. If the model that minimizes the
validation error should be chosen, <code>choose = "min"</code> should be
set. Otherwise, <code>choose = "1se"</code> leads to simplest model in the range
of one standard error of the minimizing model.</dd></dl>

Arguments

Tuning the LASSO regularization parameter — get.ideal.penalty

<dl>

<dt>model</dt>
<dd>A fitted <code>linear.logic</code> model (i.e., a model created via
<code>fitLinearLogicModel</code> or <code>fitLinearBoostingModel</code>)</dd>


<dt>X</dt>
<dd>Matrix or data frame of binary input data.
This object should correspond to the binary matrix for fitting the model.</dd>


<dt>y</dt>
<dd>Response vector. 0-1 coding for binary outcomes.</dd>


<dt>Z</dt>
<dd>Optional quantitative covariables supplied as a matrix or
data frame. Only used (and required) if the model was fitted using them.</dd>


<dt>scoring_rule</dt>
<dd>The scoring rule for evaluating the validation
error and its standard error. For classification tasks, <code>"deviance"</code>
or <code>"Brier"</code> should be used.</dd>


<dt>choose</dt>
<dd>Model selection scheme. If the model that minimizes the
validation error should be chosen, <code>choose = "min"</code> should be
set. Otherwise, <code>choose = "1se"</code> leads to simplest model in the range
of one standard error of the minimizing model.</dd>

</dl>

get.ideal.penalty: Tuning the LASSO regularization parameter

Description

Usage

Value

Arguments