D-STEM: a statistical software for modelling complex 3D and 4D space-time data
Abstract:
"Relevant physical, environmental and social phenomena usually occur at large or global scale and with a temporal dynamicd that goes from days to years. Collecting and analysing data is the only way to understand those phenomena and, when needed, to propose actions of improvement or mitigation. D-STEM is a MATLAB software that implements statistical models for analysing multivariate and functional space-time data collected at local or global scale. In this seminar, it is discussed which kind of data sets are handled by D-STEM and how the software can be used for space-time predictions and inference. D-STEM capabilities in handling large data sets are also detailed."
Referent/Referentin
Francesco Finazzi
Francesco Finazzi is Statistician and Computer Scientist at the University of Bergamo. In his talk, he will show the (statistical) background of modelling functional data. More precisely, the observed functions (profiles, trajectories etc) are allowed to be dependent in space and/or time and the dependence structure is simultaneously modelled. All parameters of this approach can be estimated using an EM algorithm based on the maximum-likelihood principle. He will also sketch, how the model is implemented in his MATLAB software D-STEM.
Termin
20. Jan. 202011:00 - 12:30
Ort
Raum: 410-411Schneiderberg 50
