# forecast.prob.MSAR: Forecast probabilities for (non) homogeneous MSAR models In NHMSAR: Non-Homogeneous Markov Switching Autoregressive Models

## Description

Computes, for each time t, the conditional probabilities for MSAR models P(Y_t|y_{1:(t-1)}) where Y is the observed process and y the observed time series.

## Usage

 `1` ```forecast.prob.MSAR(data, theta, yrange = NULL, covar.emis = NULL, covar.trans = NULL) ```

## Arguments

 `data` observed time series, array of dimension T*N.samples*d `theta` object of class MSAR including the model's parameter and description. See init.theta.MSAR for more details. `yrange` values at which to compute the forecast probabilities `covar.emis` emission covariate if any. `covar.trans` array of univariate or multivariate series of covariate to take into account in the transition probabilities. The link function is defined in the initialisation step (running init.theta.MSAR.R).

## Value

A list containing

 `..\$yrange ` abscissa for the forecast probabilities `..\$prob ` forecast probabilities `Yhat` forecasted value

## Author(s)

Valerie Monbet, [email protected]

prediction.MSAR

## Examples

 ``` 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21``` ```## Not run #data(meteo.data) #data = array(meteo.data\$temperature,c(31,41,1)) #T = dim(data)[1] #N.samples = dim(data)[2] #d = dim(data)[3] #M = 2 #theta.init = init.theta.MSAR(data,M=M,order=2,label="HH") #res.hh.2 = fit.MSAR(data,theta.init,verbose=TRUE,MaxIter=200) #FP = forecast.prob.MSAR(data,res.hh.2\$theta) #plot(data[,1,],typ="l") #lines(FP\$Yhat[,1],col="red") #alpha = .1 #IC.emp = matrix(0,2,T) #for (k in 1:length(data[,1,])) { # tmp = cumsum(FP\$prob[,k,1])/sum(FP\$prob[,k,1]) # IC.emp[1,k] = FP\$yrange[max(which(tmp

### Example output

```
```

NHMSAR documentation built on Dec. 5, 2017, 9:03 a.m.