team_name: fjordhest
model_name: ensemble
model_abbr: fjordhest-ensemble
model_contributors:
  - name: Sasi Kandula
    affiliation: Norwegian Institute of Public Health
    email: sasikiran.kandula@fhi.no
  - name: Alfonso Diz-Lois Palomares
    affiliation: Norwegian Institute of Public Health
    email: alfonso.diz-loispalomares@fhi.no
  - name: Gunnar Rø
    affiliation: Norwegian Institute of Public Health
    email: GunnarOyvindIsaksson.Ro@fhi.no
  - name: Birgitte de Blasio
    affiliation: Norwegian Institute of Public Health
    email: Birgitte.Freiesleben.DeBlasio@fhi.no
website_url: https://www.fhi.no/en/
license: cc-by-4.0
team_model_designation: primary
methods: An inverse-WIS weighted ensemble of a mechanistic model, two time series models (ARIMA, ETS) and a random walk with drift model.
data_inputs: Weekly incidence of cases, hospitalizations and deaths from the European Covid-19 forecast hub.
methods_long: For cases and deaths, an ensemble of a mechanistic model (based on EpiEstim R package), two time series models and a random walk with drift model, as implemented in the fable R package. Model weights are the inverse of the model weighted-interval scores (WIS) over the last 3 forecasts that could be evaluated. The mechanistic model for hospitalizations is a stochastic SEIR model, with a sequential Monte Carlo based inference. The estimates are solely the work of the contributors and do not necessarily represent, nor are endorsed by, the Norwegian Institute of Public Health.