partial_cv_preds

Helper function to properly format partially cross-validated predictions
from a fitted super learner.

Targeted minimum loss-based estimators of counterfactual means and
causal effects that are doubly-robust with respect both to consistency and
asymptotic normality (Benkeser et al (2017), <doi:10.1093/biomet/asx053>; MJ
van der Laan (2014), <doi:10.1515/ijb-2012-0038>).

David Benkeser

drtmle

Doubly-Robust Nonparametric Estimation and Inference

Nima Hejazi

partial_cv_preds function

<dl><dt>fit_sl</dt>
<dd>A fitted <code>SuperLearner</code> object with
<code>control$saveCVFitLibrary = TRUE</code></dd>
<dt>a_0</dt>
<dd>Treatment level to set. If <code>NULL</code>, assume this function
is being used to get partially cross-validated propensity score predictions.</dd>
<dt>W</dt>
<dd>A <code>data.frame</code> of named covariates.</dd>
<dt>family</dt>
<dd>Family of prediction model</dd>
<dt>include</dt>
<dd>A boolean vector indicating which observations were actually
used to fit the regression.</dd>
<dt>easy</dt>
<dd>A boolean indicating whether the predictions can be
computed the "easy" way, i.e., based just on the Z matrix from SuperLearner.
This is possible for propensity score models when no missing data AND no
stratification.</dd></dl>

Arguments

Helper function to properly format partially cross-validated predictions
from a fitted super learner. — partial_cv_preds

<dl>

<dt>fit_sl</dt>
<dd>A fitted <code>SuperLearner</code> object with
<code>control$saveCVFitLibrary = TRUE</code></dd>


<dt>a_0</dt>
<dd>Treatment level to set. If <code>NULL</code>, assume this function
is being used to get partially cross-validated propensity score predictions.</dd>


<dt>W</dt>
<dd>A <code>data.frame</code> of named covariates.</dd>


<dt>family</dt>
<dd>Family of prediction model</dd>


<dt>include</dt>
<dd>A boolean vector indicating which observations were actually
used to fit the regression.</dd>


<dt>easy</dt>
<dd>A boolean indicating whether the predictions can be
computed the "easy" way, i.e., based just on the Z matrix from SuperLearner.
This is possible for propensity score models when no missing data AND no
stratification.</dd>

</dl>

partial_cv_preds: Helper function to properly format partially cross-validated predictions from a fitted super learner.

Description

Usage

Arguments