getConfVectorForROC

Methods for computing ROC curves require a vector of confidences
of the positive class, while in qCBA, the confidence returned by predict.qCBARuleModel with
outputProbabilies = TRUE returns confidence for the predicted class.
This method converts the values to confidences for the positive class

Implements the Quantitative Classification-based on
Association Rules (QCBA) algorithm (<doi:10.1007/s10489-022-04370-x>).
QCBA postprocesses rule classification models making them typically smaller and in some cases more accurate.
Supported are 'CBA' implementations from 'rCBA', 'arulesCBA' and 'arc' packages, and 'CPAR', 'CMAR', 'FOIL2' and 'PRM' implementations
from 'arulesCBA' package and 'SBRL' implementation from the 'sbrl' package. The result of the post-processing is an ordered CBA-like rule list.

Tomas Kliegr

qCBA

Postprocessing of Rule Classification Models Learnt on Quantized
Data

Tomáš Kliegr

getConfVectorForROC function

<dl><dt>confidences</dt>
<dd>Vector of confidences</dd>
<dt>predictedClass</dt>
<dd>Vector with predicted classes</dd>
<dt>positiveClass</dt>
<dd>Positive class (String)</dd></dl>

Arguments

Returns vector with confidences for the positive class (useful for ROC or AUC computation) — getConfVectorForROC

<dl>

<dt>confidences</dt>
<dd>Vector of confidences</dd>


<dt>predictedClass</dt>
<dd>Vector with predicted classes</dd>


<dt>positiveClass</dt>
<dd>Positive class (String)</dd>

</dl>

getConfVectorForROC: Returns vector with confidences for the positive class (useful for ROC or AUC computation)

Description

Usage

Value

Arguments

Examples