PrInDTregAll

Regression tree based on the full sample; interpretability is checked (see 'ctestv'). 
The relationship between the target variable 'regname' and all other factor and numerical variables
in the data frame 'datain' is modeled based on all observations. 
The parameters 'conf.level', 'minsplit', and 'minbucket' can be used to control the size of the trees. 
Besides the maximal R2, the minimal MAE (Mean Absolute Error) is reported.

Optimization of conditional inference trees from the package 'party'
for classification and regression.
For optimization, the model space is searched for the best tree on the full sample by
means of repeated subsampling. Restrictions are allowed so that only trees are accepted
which do not include pre-specified uninterpretable split results (cf. Weihs & Buschfeld, 2021a).
The function PrInDT() represents the basic resampling loop for 2-class classification (cf. Weihs
& Buschfeld, 2021a). The function RePrInDT() (repeated PrInDT()) allows for repeated
applications of PrInDT() for different percentages of the observations of the large and the
small classes (cf. Weihs & Buschfeld, 2021c). The function NesPrInDT() (nested PrInDT())
allows for an extra layer of subsampling for a specific factor variable (cf. Weihs & Buschfeld,
2021b). The functions PrInDTMulev() and PrInDTMulab() deal with multilevel and multilabel
classification. In addition to these PrInDT() variants for classification, the function
PrInDTreg() has been developed for regression problems. Finally, the function PostPrInDT()
allows for a posterior analysis of the distribution of a specified variable in the terminal
nodes of a given tree.
In version 2, additionally structured sampling is implemented in functions PrInDTCstruc() and
PrInDTRstruc(). In these functions, repeated measurements data can be analyzed, too.
Moreover, multilabel 2-stage versions of classification and regression trees are
implemented in functions C2SPrInDT() and R2SPrInDT() as well as interdependent
multilabel models in functions SimCPrInDT() and SimRPrInDT(). Finally, for mixtures of
classification and regression models functions Mix2SPrInDT() and SimMixPrInDT() are
implemented. Most of these extensions of PrInDT are described in Buschfeld &
Weihs (2025Fc).
References:
-- Buschfeld, S., Weihs, C. (2025Fc) "Optimizing decision trees for the analysis of World Englishes
and sociolinguistic data", Cambridge Elements.
-- Weihs, C., Buschfeld, S. (2021a) "Combining Prediction and Interpretation in
Decision Trees (PrInDT) - a Linguistic Example" <doi:10.48550/arXiv.2103.02336>;
-- Weihs, C., Buschfeld, S. (2021b) "NesPrInDT: Nested undersampling in PrInDT"
<doi:10.48550/arXiv.2103.14931>;
-- Weihs, C., Buschfeld, S. (2021c) "Repeated undersampling in PrInDT (RePrInDT): Variation
in undersampling and prediction, and ranking of predictors in ensembles" <doi:10.48550/arXiv.2108.05129>.

Claus Weihs

PrInDT

Prediction and Interpretation in Decision Trees for
Classification and Regression

Sarah Buschfeld

Niklas Nitsch

PrInDTregAll function

<dl><dt>datain</dt>
<dd>Input data frame with class factor variable 'classname' and the 
influential variables, which need to be factors or numericals (transform logicals and character variables to factors)</dd>
<dt>regname</dt>
<dd>name of regressand variable (character)</dd>
<dt>ctestv</dt>
<dd>Vector of character strings of forbidden split results; 
(see function <code>PrInDT</code> for details.) 
If no restrictions exist, the default = NA is used.</dd>
<dt>conf.level</dt>
<dd>(1 - significance level) in function <code>ctree</code> (numerical, &gt; 0 and &lt;= 1); 
default = 0.95</dd>
<dt>minsplit</dt>
<dd>Minimum number of elements in a node to be splitted; 
default = 20</dd>
<dt>minbucket</dt>
<dd>Minimum number of elements in a node; 
default = 7</dd></dl>

Arguments

Regression tree based on the full sample; interpretability is checked (see 'ctestv'). 
The relationship between the target variable 'regname' and all other factor and numerical variables
in the data frame 'datain' is modeled based on all observations. 
The parameters 'conf.level', 'minsplit', and 'minbucket' can be used to control the size of the trees. 
Besides the maximal R2, the minimal MAE (Mean Absolute Error) is reported.

Regression tree based on all observations — PrInDTregAll

<dl>

<dt>datain</dt>
<dd>Input data frame with class factor variable 'classname' and the 
influential variables, which need to be factors or numericals (transform logicals and character variables to factors)</dd>


<dt>regname</dt>
<dd>name of regressand variable (character)</dd>


<dt>ctestv</dt>
<dd>Vector of character strings of forbidden split results; 
(see function <code>PrInDT</code> for details.) 
If no restrictions exist, the default = NA is used.</dd>


<dt>conf.level</dt>
<dd>(1 - significance level) in function <code>ctree</code> (numerical, &gt; 0 and &lt;= 1); 
default = 0.95</dd>


<dt>minsplit</dt>
<dd>Minimum number of elements in a node to be splitted; 
default = 20</dd>


<dt>minbucket</dt>
<dd>Minimum number of elements in a node; 
default = 7</dd>

</dl>

PrInDTregAll: Regression tree based on all observations

Description

Usage

Value

Arguments

Details

Examples