Un modelo vectorial autorregresivo para series irregularmente espaciadas no estacionarias
...
Molina Díaz, Oscar Emilio | 2014-06-26
In this work we propose a vector autoregressive model for irregularly spaced nonstationary
time series. Initially we propose this model for two time series with autoregressive
order 1 and then this model is generalized for m time series with autoregressive
order p > 1. The functional parameters of the autoregressive matrices are expanded in
a wavelet basis in order to allow the estimation by least squares after truncation at a
suitable resolution level. Besides some properties of the estimator are analyzed.
This model is applied to five irregularly spaced non-stationary biological time series
from an ecological study accomplished on the water column at the facility station Paraná
located in the Beagle Channel, Tierra de Fuego (Argentina), where is desired to
know the dependence relationships among them.
Keywords: Irregularly time series, autoregressive models, wavelets, least squares, functional
parameters.
LEER