|
Description: |
The Kalman Filter (KF), initially submitted by the Hungarian Rudolf E. Kalman in the context of control problems, has been widely applied in statistical modelling of stochastic processes that allow a state space representation. The success of the KF is attributed to the optimal estimators that it provides (in terms of linearity and Gaussian errors) as well as the fact that the equations are in a recursive form allowing an online estimation of unobservable variables, the states. However, the implementation of the KF involves the specification of the state space model and other problems arise such as the estimation of parameters, in addition to the verification of some assumptions. These issues will be addressed in particular the representation in state space of some common and useful models in applications, the estimation by the method of maximum likelihood and distribution-free estimators and the optimality conditions of the KF predictors.
Examples with environmental data sets will be given in order to illustrate the issues presented; for instance, in the area rainfall estimation or in the surface water quality monitoring.
|
Date: |
|
Start Time: |
14:30 |
Speaker: |
Marco Costa (ESTGA & CIDMA, Univ. Aveiro)
|
Institution: |
ESTGA & CIDMA, Univ. Aveiro
|
Place: |
Sala 5.5
|
Research Groups: |
-Probability and Statistics
|
See more:
|
<Main>
|
|