Classification trees for preference data: a distance-based approach

Risultato della ricerca: Paper

Abstract

In the framework of preference rankings, when the interest lies in explaining which predictors and which interactions among predictors are able to explain the observed preference structures, the possibility to derive consensus measures using a classi cation tree represents a novelty and an important tool given its easy interpretability. In this work we propose the use of a multivariate decision tree where a weighted Kemeny distance is used both to evaluate the distances between rankings and to de ne an impurity measure to be used in the recursive partitioning. The proposed approach allows also to weight di erently high distances in rankings in the top and in the bottom alternatives.
Lingua originaleEnglish
Stato di pubblicazionePublished - 2014

Fingerprint

Ranking
Predictors
Preference structure
Interaction
Novelty
Partitioning
Impurities

Cita questo

@conference{da796728e4684e679e32bfeeda8f1404,
title = "Classification trees for preference data: a distance-based approach",
abstract = "In the framework of preference rankings, when the interest lies in explaining which predictors and which interactions among predictors are able to explain the observed preference structures, the possibility to derive consensus measures using a classi cation tree represents a novelty and an important tool given its easy interpretability. In this work we propose the use of a multivariate decision tree where a weighted Kemeny distance is used both to evaluate the distances between rankings and to de ne an impurity measure to be used in the recursive partitioning. The proposed approach allows also to weight di erently high distances in rankings in the top and in the bottom alternatives.",
author = "Antonella Plaia and Mariangela Sciandra",
year = "2014",
language = "English",

}

TY - CONF

T1 - Classification trees for preference data: a distance-based approach

AU - Plaia, Antonella

AU - Sciandra, Mariangela

PY - 2014

Y1 - 2014

N2 - In the framework of preference rankings, when the interest lies in explaining which predictors and which interactions among predictors are able to explain the observed preference structures, the possibility to derive consensus measures using a classi cation tree represents a novelty and an important tool given its easy interpretability. In this work we propose the use of a multivariate decision tree where a weighted Kemeny distance is used both to evaluate the distances between rankings and to de ne an impurity measure to be used in the recursive partitioning. The proposed approach allows also to weight di erently high distances in rankings in the top and in the bottom alternatives.

AB - In the framework of preference rankings, when the interest lies in explaining which predictors and which interactions among predictors are able to explain the observed preference structures, the possibility to derive consensus measures using a classi cation tree represents a novelty and an important tool given its easy interpretability. In this work we propose the use of a multivariate decision tree where a weighted Kemeny distance is used both to evaluate the distances between rankings and to de ne an impurity measure to be used in the recursive partitioning. The proposed approach allows also to weight di erently high distances in rankings in the top and in the bottom alternatives.

UR - http://hdl.handle.net/10447/96229

UR - http://www.statmod.org/cgi-bin/getfile.pl?f=proceedings/iwsm2014_proceedings_vol2.pdf

M3 - Paper

ER -