mlpm {LPM}

Multivariate Linear Parametric Model

Description

Multivariate modelling using VAR(p) and SVAR(p) different estimation methods, simulation, daily rainfall simulated series correction and deseasonalization are performed

Usage

mlpm(x, p, prob, nsim, smean, svar, fre = 365, outer = 0,plot = F, 
rain = T, over = T, estimate = "ols", CCFlag = 20, Plag = 20, lsign = 0.05, des = T)

Arguments

x

Multivariate series

p

Model order

prob

Condifidence interval used to fix parameters in SVAR(p) model

nsim

Number of series to simulated

smean, svar

Mean and Variance smoothing windows of STL modified method

fre

Series frequency. Default fre = 365

outer

Outer loops of STL modified method. Default outer = 0

plot

Logical variable: T to receive some graphics. Default plot = F

rain

Logical variable: T to apply rain adaptor to simulated series. Default rain = F

over

Logical variable: T to use SVAR(p) model estimated with EGLS method. Need estimate = 'ols' Default over = T

estimate

Define VAR(p) estimation method. 'ols', 'burg', 'yw' (Yule-Walker). Default estimate = 'ols'

CCFlag

Lag of (Partial) Auto-CrossCorrelation function graphics . Default CCFlag = 20

Plag

Maximum lag of A-CCF used in the Portmanteau Test. Default Plag = 20

lsign

Portmanteau Test significance level. Default lsign = 0.05

des

Logical variable: T to remove seasonal components

Details

Need integer periodical datasets. Simulation use Lutkepohl algorithm with a residuals vectorial permutation to obtain innovations. Parameters selections of EGLS method is defined by t-ratio approach.

Value

coeff

List of estimated coefficients matrix

coeffstd

List of estimated standard deviations coefficients matrix. Only for OLS and EGLS method

struct

List of 'structure' of SVAR(p) model (1 define position of estimated parameter). Only for EGLS method

res

Residual series

fit

Output List of ar function

aic

Akaike Information Criterion index

Q

Portmonteau statistic

sim

List of simulated series

Note

Portmonteau test, AIC e SBC index are displaied during application. Portmonteau test is positive if Q < chi square.

Author(s)

Corrado Tallerini

References

Grimaldi S., Tallerini C., Serinaldi F. (2004) 'Modelli multivariati lineari per la generazione di serie di precipitazioni giornaliere' Giornata di Studio: Metodi Statistici e Matematici per l'Analisi Idrologiche Napoli 2004

Grimaldi S. , Serinaldi F. & Tallerini C. (2004) 'Multivariate linear parametric models applied to daily rainfall time series' Mediterranean Storms, 6rd EGU Plinius Conference held in Mediterranean Sea, Italy, October 2004

Lutkepohl, H. (1993) Introduction to Multiple Time Series Analysis 2nd edition, Springer-Verlag, Berlin.

Grimaldi, S., 'Linear parametric models applied on daily hydrological series', Journal of Hydrologic Engineering, Vol. 9, No 5 , September 2004.

Brockwell, P.J and Davis, R.A. (1990) Time Series: Theory and Methods 2nd edition, Springer, NY.

Hipel, K.W. and McLeod, A.I., (1994) Time Series Modelling of Water Resources and Enviromental Systems, Reading, UK.

Hosking, J.R.M. (1980) 'The Multivariate Portmanteau Statistic' Journal of the American Statistical Association, Vol.75, N.371, 502-608.

See Also

lpm, ar.egls, rain.adapt

Examples

##-- Mrain=mlpm(series.rainfall,3,0.95,0,120,120)
##-- Apply a SVAR(3) model with selection probability 95 % to series.rainfall
##-- after preventive deseasonalization with smoothing 120.