Title: Recursive Partitioning and related methods
Description:
Various nonparametric methods have been
introduced to solve real problems. The most appealing aspect for the recursive
partitioning user is the final tree that provides a comprehensive description of
the phenomenon in different frameworks such as marketing, credit scoring,
finance, medical diagnosis, etc. As a matter of fact, users very often accept
statistical results only if these confirm theoretical hypotheses on the
phenomenon derived from prior knowledge. Thus, several open questions arise when
using such heuristic tools; in recursive partitioning the most difficult
pertains to which tree to consider for explaining the dependence data structure
and how to evaluate the accuracy of the final tree classifier if this should be
extended to unseen objects without considering any ¡Èinferential dogma¡É. This
latter aspect makes recursive partitioning methods to be considered both as an
exploratory tool and as a confirmatory nonparametric model, also known as
decision rule. A distinction is made between whether to explore dependency (exploraion)
or to predict and decide about future responses on the basis of the selected
predictors (confirmation).
We will cover new statistical and computational topics in recursive partitioning
emphasizing the use of tree based models for exploratory and confirmative goals,
as well as discussing alternatives to standard recursive partitioning algorithms
and presenting applications on real complex data.
Focus:
Tree-based methods
Non-standard data
Data editing
Data mining
Interaction effects models
Ensemble classifiers
Co-Chairs:
Roberta Siciliano
Dipartimento di Matematica e Statistica
University of Naples Federico II
Via Cintia, Monte S. Angelo
I-80126, Napoli ITALY
Tel: +39 081/675120
Fax: +39 081/675109
Email: roberta@unina.it
|
Jacqueline J. Meulman
P.O. Box 9555, Pieter de la Court building
2300 RB Leiden,
Wassenaarseweg 52,The Netherlands
Tel: +31-71-527 4105
Fax: +31-71-527 3865
Email: meulman@fsw.leidenuniv.nl |
Claudio Conversano
Department of Economics
University of Cassino
Via Mazzaroppi, Cassino(FR),
ITALY I-03043
Tel: +39 0776 2993448
Fax: +39 0776 310892
Email: c.conversano@unicas.it |
|
|