inject_uncertainty

Adds stochastic noise to point predictions to properly reflect
imputation uncertainty. Called after model prediction, before
value assignment.

internal

Provides methods for imputation and visualization of
missing values. It includes graphical tools to explore the amount, structure
and patterns of missing and/or imputed values, supporting exploratory
data analysis and helping to investigate potential missingness mechanisms
(details in Alfons, Templ and Filzmoser, <doi:10.1007/s11634-011-0102-y>).
The quality of imputations can be assessed visually using a wide range of
univariate, bivariate and multivariate plots.
The package further provides several imputation methods,
including efficient implementations of k-nearest neighbour and hot-deck
imputation (Kowarik and Templ 2013, <doi:10.18637/jss.v074.i07>),
iterative robust model-based multiple
imputation (Templ 2011, <doi:10.1016/j.csda.2011.04.012>;
Templ 2023, <doi:10.3390/math11122729>), and machine learning–based
approaches such as robust GAM-based multiple imputation
(Templ 2024, <doi:10.1007/s11222-024-10429-1>) as well as random forest
and gradient boosting (XGBoost) imputation
(Niederhametner et al., <doi:10.1177/18747655251339401>).
General background and practical guidance on imputation are provided in the
Springer book by
Templ (2023) <doi:10.1007/978-3-031-30073-8>.

Matthias Templ

Visualization and Imputation of Missing Values

Alexander Kowarik

Andreas Alfons

Johannes Gussenbauer

Nina Niederhametner

Eileen Vattheuer

Gregor de Cillia

Bernd Prantner

Wolfgang Rannetbauer

inject_uncertainty function

<dl><dt>preds</dt>
<dd>Numeric vector of predicted values for missing observations</dd>
<dt>method</dt>
<dd>One of "none", "normalerror", "resid", "pmm", "midastouch"</dd>
<dt>scale</dt>
<dd>Scale estimate (sigma hat) from model. Required for "normalerror".</dd>
<dt>residuals</dt>
<dd>Training residuals. Required for "resid".</dd>
<dt>y_obs</dt>
<dd>Observed values of target variable. Required for "pmm", "midastouch".</dd>
<dt>score_obs</dt>
<dd>Model scores for observed rows. Required for "pmm".</dd>
<dt>score_miss</dt>
<dd>Model scores for missing rows. Required for "pmm".</dd>
<dt>pmm_k</dt>
<dd>Number of donors for pmm/midastouch (default 5).</dd>
<dt>pmm_k_method</dt>
<dd>Aggregation for pmm when k &gt; 1 (default "random").</dd>
<dt>X_obs</dt>
<dd>Predictor matrix for observed rows. Required for "midastouch".</dd>
<dt>X_miss</dt>
<dd>Predictor matrix for missing rows. Required for "midastouch".</dd></dl>

Arguments

Inject imputation uncertainty into predictions — inject_uncertainty

<dl>

<dt>preds</dt>
<dd>Numeric vector of predicted values for missing observations</dd>


<dt>method</dt>
<dd>One of "none", "normalerror", "resid", "pmm", "midastouch"</dd>


<dt>scale</dt>
<dd>Scale estimate (sigma hat) from model. Required for "normalerror".</dd>


<dt>residuals</dt>
<dd>Training residuals. Required for "resid".</dd>


<dt>y_obs</dt>
<dd>Observed values of target variable. Required for "pmm", "midastouch".</dd>


<dt>score_obs</dt>
<dd>Model scores for observed rows. Required for "pmm".</dd>


<dt>score_miss</dt>
<dd>Model scores for missing rows. Required for "pmm".</dd>


<dt>pmm_k</dt>
<dd>Number of donors for pmm/midastouch (default 5).</dd>


<dt>pmm_k_method</dt>
<dd>Aggregation for pmm when k &gt; 1 (default "random").</dd>


<dt>X_obs</dt>
<dd>Predictor matrix for observed rows. Required for "midastouch".</dd>


<dt>X_miss</dt>
<dd>Predictor matrix for missing rows. Required for "midastouch".</dd>

</dl>

Inject imputation uncertainty into predictions

inject_uncertainty: Inject imputation uncertainty into predictions

Description

Usage

Value

Arguments