lps {BGVAR}R Documentation

Log predictive scores (LPS)

Description

Calculates log predictive scores using the normal distribution as approximation for the predictive density. The function calculates LPS for the marginal densities and not for the joint predictive distribution (of all variables in the GVAR system). If the forecast horizon is greater than 1 (fhorz in fcast), LPS is the sum of single LPS over the forecast horizon per variable.

Usage

lps(fcastM)

Arguments

fcastM

An object fitted using fcast. Make sure that you have set h>0 in fcast.

Value

lps.scores

A vector that returns the log predictive scores.

Author(s)

Martin Feldkircher

See Also

See also fcast

Examples

library(BGVAR)
data(monthlyData)
model.sims<-bgvar(Data=new.data,W=W,saves=100,burns=100,plag=1,hyperparm=NULL,Cpu=1,prior="SIMS",save_thin=1,eigen=TRUE,logfile=FALSE,stats=TRUE)
fcastM<-fcast(model.sims,h=5,fhorz=8)

lps(fcastM)

[Package BGVAR version 1.1.3 Index]