Ph.D. Programme

Statistical analysis and computational statistics

The Statistics programme offers modern statistical thinking and tools for solving problems in various applied fields. The main object is analysis of data consisting of dependent observations. The research areas are analysis of spatial data, computationally intensive statistical methods, analysis of data obtained through complex sampling, analysis of longitudinal data and industrial statistics.

Head of the programme

Prof. Antti Penttinen

Department of Statistics

ojahannu@cc.jyu.fi

Programme information

Robust and nonparametric multivariate methods

Supervisors: Prof. Hannu Oja, Dr. Annaliisa Kankainen

The theory and algorithms for robust and nonparametric multivariate inference (testing and estimation in the MANOVA, principal component analysis, multivariate regression, canonical analysis, discriminant and cluster analysis problems) are developed. The main emphasis is on methods based on multivariate signs and ranks. Project homepage

Statistical analysis of spatial data

Supervisor: Prof. Antti Penttinen

This program consists of method development and applications for statistical analysis of spatial data and for dependent data in general. The approaches employed are marked point process modelling, geostatistics, hierarchical Bayesian modelling, stochastic geometry, stereology and statistical image analysis. Simulation-based statistical inference including Markov chain Monte Carlo (MCMC) methodology has a central role. Areas of application are biology, forestry, medical science, epidemiology, materials science and industrial statistics.

Survey sampling

Supervisors: Prof. Risto Lehtonen, Dr Ari Veijanen, Prof. Carl-Erik Särndal.

This area contains research in model-assisted survey sampling including the generalized regression (GREG) estimators in the context of estimation for population domains and small areas. Generalized linear mixed models are examined as potential assisting models in the GREG estimators. The relative behavior (bias and accuracy) of the constructed estimators is examined theoretically and by simulation techniques. The applications are in official statistics and related fields, and the program is co-operating with Statistics Finland.

Multivariate Statistics

Supervisor: Prof. Esko Leskinen.

Multivariate statistics contains development and applications of structural equation models in multidiciplinary research. The main emphasis in the analysis of longitudinal data, the measurement theory and the applications of LISREL models. The applications are in health sciences, gerontology and psychology.

Back to COMAS main page