The
Ninth Course in the ECAS Programme:
Data Mining and
Explorative Multivariate Data Analysis |
SAN MARCO DI
CASTELLABATE, ITALY |
September
28 - October 4, 2003 |
Scientific
Programme Committee
H.H. Bock, University of Aachen, Germany
L. D´Ambra, University of
Naples, Italy (Chair)
B. Fichet, University of Marseilles,
France
W. Krazanowski, School of Math.
Sciences, Exeter, England
D. Peña, Universidad Carlos
III, Madrid, Spain
M. Vichi, University of Rome, Italy
Organising
Committee
P. Amenta, University of Lecce
L. D´Ambra, University of
Naples (Chair)
R. Lombardo, Second University of Naples
P. Sarnacchiaro, University of Naples
B. Simonetti, University of Naples
Scope
of the course
In the context of data mining
process,
one of the goal is to aggregate or amalgamate
the information contained in large data-sets into manageable (smaller)
information nuggets.
Data reduction methods can include sophisticated techniques like
clustering, principal component
analysis, etc... However, an important general difference in the focus
and purpose between data mining and traditional exploratory data
analysis is that data mining is more oriented towards
applications than less concerned with identifying the basic nature on
the underlying phenomena.
Multivariate exploratory techniques, designed specifically to identify
patterns in multivariate data sets, include: clusters analysis and
classification trees, linear and non-linear principal component
analysis, multidimensional scaling, stepwise linear and non-linear
regression, principal component regression, partial least squares,
etc...
The course aims to show how different problems from several domains can
be fruitfully tackled with methods from multivariate data analysis
which are capable of providing specialized tools for treating complex
data in data mining process.
To further facilitate the understanding of multivariate analysis,
strong importance will be given to applications
from different types of problems which underline the benefits of
multivariate analysis.
Topics
- Data Mining: complexity, cleaning compression.
Definition and objectives.
R.D. De Veaux (Williams college, USA)
- Data Mining: complexity, visualizing low and
high dimensionnal data.
R.D. De Veaux (Williams college, USA)
- Multivariate explorative data analysis: the
linear and non-linear approaches through splines
J.F. Durand (University of Montpellier, France)
- Multivariate regression analysis wiyh strongly
correlated variables. Linear partial least squares and structural
equation modelling
G. Vittadini (University of Milan, Italy)
- Classification trees, random forests, MART and
MARS as variations on trees.
R.D. De Veaux (Williams college, USA)
- Methods for assessing reliability of outcomes
and determining numbers of components
H. Kiers (University of Groningen, The Netherlands)
- Multivariate regression analysis with strongly
correlated variables. Reduced rank regression. Principal covariates
regression. Principal components regressions
H. Kiers (University of Groningen, The Netherlands)
- Classification and clustering
M. Vichi and H. Bock (University of Rome, Italy, and University of
Aachen, Germany)
- Neural networks and data mining
M. Vichi and H. Bock (University of Rome, Italy, and University of
Aachen, Germany)
- Perspective and conclusion on data mining and
explorative multivariate data analysis
H. Kiers (University of Groningen, The Netherlands)