Tag: probabilistic forecasting

Minimum CRPS vs. maximum likelihood

news

In a new paper in Monthly Weather Review, minimum CRPS and maximum likelihood estimation are compared for fitting heteroscedastic (or nonhomogenous) regression models under different response distributions. Minimum CRPS is more robust to distributional misspecification while maximum likelihood is slightly more efficient under correct specification. An R implementation is available in the crch package. Read more ›

Thunderstorm forecasting with GAMs

news

Boosted binary generalized additive models (GAMs) with stability selection and corresponding MCMC-based credibility intervals are discussed in a new MWR paper as a probabilistic forecasting method for the occurrence of thunderstorms. Read more ›