ensembleBMA (version 2.0)

crpsANDmae: CRPS and MAE via sampling

Description

Computes the continuous ranked probability score and median absolute error for ensemble BMA forecasting models via sampling. Can be used for distributions for which the CDF is not available.

Usage

crpsANDmae( object, ensembleData, nSamples = 10000, seed = NULL, 
            popData = NULL, ...)

Arguments

object

An ensemble BMA model fit for ensembleData.

ensembleData

An ensembleData object giving including ensemble forecasts, observations and dates corresponding to object.

nSamples

The number of simulation samples (gamma0 for precipitation model only)

seed

Argument to set.seed for random number generation in simulation.

popData

For ensembleBMAgamma0 and fitBMAgamma0 objects, there is an additional popData argument for providing predictors in the logistic regression for probability of precipitation. The value of popData

...

Included for generic function compatibility.

Value

• A matrix whose rows correspond to the crps and mae, respectively, and whose columns correpond to these values for climatology, the ensemble, the BMA model, and simulations from the BMA model.

Details

This method is generic, and can be applied to all ensemble BMA forecasting models. Note the model may have been applied to a transformation of the data, but that information is included in the input x, and the output is transformed appropriately. For the gamma0 model for precipitation, the CRPS is only available through simulation, because a transformation of the forecast values is being modeled rather than the forecast values themselves (the default transformation is the cube root). For normal models for temperature and pressure, simulation is not need and the nSamples and seed arguments are ignored.

References

D. A. Unger, A method to estimate the continuous ranked probability score, Preprints of the Ninth Conference on Probability and Statistics in Atmospheric Sciences, Virginia Beach, VA USA, 206-213, American Meteorological Society.

H. Hersbach, Decomposition of the continuous ranked probability score for ensemble prediction systems, Weather and Forecasting 15, 559-570.

T. Gneiting and A. E. Raftery, Strictly proper scoring rules, prediction and estimation, Technical Report No. 463R, Department of Statistics, University of Washington, November 2006 (to appear in Journal of the American Statistical Association).

See Also

ensembleBMA

Examples

data(slp)
  
  slpData <- ensembleData(forecasts = slp[c("AVN","GEM","ETA","NGM","NOGAPS")],
                          observations = slp$obs, dates = slp$date)

  slpFit <- ensembleBMAnormal(slpData)

  crpsANDmae( slpFit, slpData)