Time series and spatial analysis

  • Aide
  • Recherche
  • Facebook
  • Twitter
UFR Sciences et techniques de la Côte Basque

Time series and spatial analysis

Présentation

Course description:
Space and time are often vital components of ecological data sets. Accounting for the space and time information in statistical models is beneficial when the response variable in question is proved to have a space and time dependence. 
This course unit focuses on methods for spatial analysis and time series using the R statistics software. The tutorials illustrate many analyses using ecological data. Informations on how to use some R packages dedicated to these analyses are also presented.

 

Handling and visualizing spatio-temporal Data

Creating maps with R

 Understanding and reducing high-dimensional spatio-temporal data:

  • Graphical displays of spatio-temporal data
  • PLS  methods (classical and sparse)
  • Multivariate datamining methods for the analysis of one, two or more datasets simultaneously (PCA, CCA, Within and between PCA, RLQ method, multivariate spatial analysis involving neighborhood)

Modeling spatio-temporal data in ecology:

  • PLS Regression
  • Bayesian Modeling: The course is a first and preliminary introduction to Bayesian modeling for applied aquatic ecology. It is based on a real case study (estimation of abundance by successive removals) which data have been collected in the field.

Software used : R and associated specific packages

Course structure:

 

Time series (15h00)

Datamining methods (17h00)

Modelling spatio-temporal data (18h00)

Time series
1- Characteristics of Time Series
   -The Nature of Time Series Data
   -Some simple forecasting methods
2- Time domain Model: AR, MA and ARIMA models
   -AR model
   -MA model
   -ARIMA model
3- Decomposition methods
   -Multiplicative and additive models
   -Moving average method and STL method
   -Forecast
4- Exponential smoothing
   -Simple exponential smoothing
   -Holt’s linear trend method
   -Damped trend methods

  • For one dataset: PCA and sparse PCA
  • For 2 datasets: CCA, PLS analysis
  • More datasets: RLQ Analysis,between-within PCA or CA.
  • Introducing neighborhood into multivariate analysis :
  • Geostatistical interpolation: variogram and kriging
 
  • GLM: Poisson, logistic and general formulation
Bayesian modeling:introduction of some basic concepts and theory of Bayesian modeling. Monte Carlo Markov Chain (MCMC) sampling techniques, by means of the (Open)BUGS software, are also introduced as key numerical tools for the practice of Bayesian modeling.

 

ID

INTENDED LEARNING OUTCOMES

LEVEL

(*)

 

Handling and visualizing spatio-temporal Data

AP

 

Understanding and reducing high-dimensional spatio-temporal data

AN

 

Modeling spatio-temporal data in ecology

AP

 (*) Levels: K: Knowledge, C: Comprehension, AP: Application, AN: Analysis, S: Synthesis, E: Expertness

 

Conditions d'admission

Prerequisites : Univariate and Multivariate Linear model, R software

Examens

written exams and project evaluation using posters

En bref

Crédits ECTS 5

Nombre d'heures 50

Niveau d'étude BAC +5

Contact(s)

Composante

Responsable(s)

Bru Noëlle

Collège STEE Sciences et Techniques ANGLET
1 Allée du Parc Montaury
64600 ANGLET
Tél : +33 559574496

Email : noelle.bru @ univ-pau.fr

Lieu(x)

  • Anglet