Catherine Paquin's Research Page |
||
/main page/research axes/Catherine Paquin |
||
Former Position |
Graduate
student M.Sc. | 2017 - 2018 |
|
Contact |
Engineer Norda Stelo Montreal, CANADA |
|
Education |
M.Sc., Civil Engineering |
2016-2018 |
|
Research Project |
Title: Modelling Environemental Effect Dependencies with Principal Component Analysis in Bayesian Dynamic Linear Models Summary:
My master thesis proposes a new method adapted to BDLM that
uses PCA on environmental effect observations in order to model the
dependency between environmental effects and structural response. The
proposed method is to model the linear dependency between the
structural response model and the individual impacts of the
environmental effects models transformed in an un- correlated space
obtained through PCA. This modification to BDLM only modifies a model
matrix and does not change the method’s essence. The case-study results
show that the method is able to handle
multiple correlated environmental effects in BDLM, which was not
possible before, and that including in a structural response model the
information from multiple environmental effect datasets leads to an
increased prediction capacity. The method could also be expanded to
handle different types of environmental effects at once by adding a
pre-processing step. |
|
Publications |
[2018] Modelling Environemental Effect Dependencies with Principal Component Analysis in Bayesian Dynamic Linear Models Paquin, C. M.Sc. Thesis, Polytechnique Montréal [PDF]
|