David NERINI
e-mail : david.nerini@univmed.fr
UMR LMGEM 6117 CNRS
Campus de Luminy, Case
901
Telephone: (+33)
491 829 125
Fax: (+33) 491 829 119
location: Bat. TPR1, Entree
F, porte 442
Welcome
to David Nerini's
home page. I hold an assistant professor
position at the Laboratoire de
Microbiologie, Geochimie
et Ecologie Marines in the Centre d'Oceanologie
de Marseille
I
teach courses on statistics and computer sciences
for life sciences at the senior undergraduate and graduate level. Free courses will
be soon
available hereunder :
· Regression & Approximation
· Time series & Stochastic Processes
· Introduction to statistics & probabilities I, first year courses
· Data-Mining, France CO2 emissions data
·
Regression
Tree for
functional data analysis
The
analysis of relationships between a functional response variable and
a set of real predictors through regression tree methods (Nerini
& Ghattas,
2007).
Badih Ghattas,
and I have developed a set of functions in R for the construction
of a
functional regression tree model.
(Coll. : Badih Ghattas)
· Analysis
of shape for
functional objects
Otoliths are small pieces of calcium carbonate located into the inner ear of Teleostean fishes. The study of their shapes is a good indicator for stock discrimination and provides nice informations about environmental conditions. However, the study of these closed contours must be carefully proceeded for several reasons. Otoliths can be seen as parametrized closed curves and the comparison of a collection of such objects must be achieved using methods which allow invariance under any linear transformations (translation, rotation, dilatation).We propose the use of tools developed for functional data analysis (Ramsay & Silverman, 2005) applied in this Procrustes context, in order to classify a collection of otoliths and to identify the different sources of shape variability between individuals of same species. Download slides. Download French slides (Jan. 2008, more complete) Draft available soon.
(Coll. : Fabien Morat, Yves Letourneur, UMR DIMAR, Centre d'Océanologie de Marseille)
·
Fractional
Brownian Motion (FBM)
in chemostat
A
chemostat
is a device for the continuous growth of microorganisms (phytoplankton,
bacteria, …)
in which growth rate
and population size can be regulated by the concentration of a limiting
nutrient. Starting with four experiences realized with periodic input
of
nutrient, estimations of the delay between the input and the response
of the
population are achieved with standard Fourier analysis. Once the
deterministic part of the system has been isolated by regression, we
propose a statistical model to study the sructure of the residuals. We
show that, in all cases, the noise
structure is close to a FBM with the same Hurst parameter.
Link toward pdf file.
(Coll. : Claude Manté, J-C Poggiale)
· Zooplankton
Size Spectra
Analysis:
Observations in a chemostat can be achieved with an optical counter. We propose methods in order to analyse such data.
(Coll. : Antoine Sciandra*, Olivier Bernard**, J-C Poggiale***, C. Manté***, F. Carlotti***)
* : Laboratoire d'Océanographie Villefranche, ** : INRIA, Sophia-Antipolis, *** : COM, Marseille
· R program for Electrophoresis PCA
Data analysis for electrophoresis spectra. We propose an aggregation method which allows to use PCA with respect to the functional structure of the spectra. Draft available soon.
Linear Kriging for functional data
In
oceanography, we often deal with data that can be considered as curves.
We propose a method which allows to map functional data taking into
account of the spatial structure of the sampled curves. Links toward the proposed summary for the VIII International Geostatistics Congress and proposed long summary for the First International
Workshop on Functional and Operatorial Statistics , IWFOS2008, June 2008, Toulouse.
Recent Interesting Manuscripts:
· D. Nerini, Morat, F. & L., Letourneur, (2007). Procustes Shape Analysis of Functional Data. Draft soon available
· D. Nerini and Ghattas, B., (2007). Functional data analysis with regression trees. Draft.
·
D. Nerini
and Ghattas G., (2007).
Regression tree for functional response
: applications in oceanology. Comp. Stat. & Data Analysis.
· Poggiale,
J.-C., Auger, P., Nerini,
D.,Manté,
C., Gilbert, F., (2005). Global
production increased by spatial heterogeneity in a
prey-predator model.
Acta
Biotheoretica
(52) : 369-380.