Universidad Complutense de Madrid
E-Prints Complutense

A non-linear model applied to the precipitation time series of San Fernando, Cadiz

Impacto

Downloads

Downloads per month over past year



Escobar, Daniela and Fenoy, Mar and Luna, M. Y. and Martin, M. A. and Morata, A. and Pascual, A. (2014) A non-linear model applied to the precipitation time series of San Fernando, Cadiz. In International work-conference on time series. Copicentro, Granada, pp. 793-803. ISBN 978-84-15814-97-9



Abstract

The longest time series of precipitation in San Fernando, Cadiz will be analized. The avaiable data is from January of 1851 to December of 2008. The goal is to fit a model and estimate the parameters in order to predict future values of the time series. It is noted that the time series do not have a tendency but it has a seasonal component. A first study fitting a Holt-Winters model is made. In order to fit an ARIMA model it is neccesary to achieve a stationary process. Finally there is a discussion wether to use a non-linear model since the time series presents many atipical points and high curtosis. The residuals of the ARIMA model are fitted for a non-linear model GARCH since they are uncorrelated but dependent.


Item Type:Book Section
Additional Information:

1st International Work-Conference on Time Series (ITISE)

Uncontrolled Keywords:Time Series; Stationary Process; Heteroscedasticity
Subjects:Sciences > Mathematics > Mathematical statistics
ID Code:33254
Deposited On:24 Sep 2015 06:45
Last Modified:24 Sep 2015 06:45

Origin of downloads

Repository Staff Only: item control page