GetDensity {SpecsVerification}

Calculate density and integrated density function of a dressed ensemble forecast at a matrix of values

Description

Calculate density and integrated density function of a dressed ensemble forecast at a matrix of values

Usage

GetDensity(dressed.ens, x, integrated = FALSE)

Arguments

dressed.ens

A list returned by the function 'DressEnsemble'. See '?DressEnsemble' for details.

x

A matrix with either 1 row or nrow(dressed.ens[["ens"]]) rows and an arbitrary number of columns, holding the arguments at which the forecast distributions are to be evaluated. See Details.

integrated

logical, (default=FALSE): If 'integrated' is TRUE, the integrated density (i.e. the value of the cumulative distribution function) is returned, otherwise the value of the density is returned.

Details

If you want to evaluate each forecast distribution function at the same x-values, a matrix with one row can be provided, e.g. 'x = matrix(c(-1, 0, 1), nrow=1)'

If the N individual forecast distributions are to be evaluated at different x-values, a matrix with N rows must be provided, where N is the number of time instances.

To calculate the PIT values for the dressed ensemble and observations 'obs', use 'GetDensity(dressed.ens, x = matrix(obs, ncol=1), integrated=TRUE)'

Value

The function returns a matrix, whose rows correspond to the individual ensemble forecasts and whose columns correspond to the values provided by the argument 'x'.

See Also

DressEnsemble, DressCrps, DressIgn, PlotDressedEns, FitAkdParameters

Examples

 data(eurotempforecast)
 dressed.ens <- DressEnsemble(ens)
 # calculate each density at the same x-values
 x1 <- matrix(seq(-3, 3, 0.1), nrow=1)
 dens1 <- GetDensity(dressed.ens, x1)
 # get the densities that the forecast 
 # distributions assign to the observations
 x2 <- matrix(obs, ncol=1)
 dens2 <- GetDensity(dressed.ens, x2)
 # get the integrated densities that the forecast 
 # distributions assign to the observations (useful
 # for constructing a PIT histogram)
 pit <- GetDensity(dressed.ens, x2, integrated=TRUE)