code

Guide to code

load provides functions to access different types of European Forecast Hub data, with example scripts for using those functions to create a dataset of forecasts or evaluations. summarise creates the summary numbers and figures used in the main study.

load

Complete scripts

• load-and-process-forecasts.R - script to download forecasts and remove anomalies
• evaluation-scores.R - script to load and save evaluation scores

Functions

Scripts containing only functions (package-style), including the type of output they return in (brackets):

• Functions to access forecasts

  • download_forecasts.R
    • download_model_forecasts() - get forecasts from a single model at specified forecast dates (dataframe)
    • download_forecasts() - get forecasts from optionally specified models at optionally specified dates (dataframe)
    • process_forecasts() - add observed data and remove anomalies (dataframe)
  • download_variant_forecasts.R
    • download_variant_forecasts() - gets Hub ensemble forecast at timing of variant introductions (dataframe)

• Functions to access metadata

  • download_metadata.R - collection of helper functions to access metadata about the Forecast Hub, including:
    • download_hub_metadata() - get information about the Hub configuration, such as start date, targets (dataframe)
    • get_forecast_dates() - get all possible weekly dates on which forecasts were made (vector)
    • download_model_names() - get names of all forecasting models in the Hub (vector)
    • download_model_metadata() - get metadata associated with all models in the Hub (dataframe)
    • download_anomalies() - download dataset of anomalies found by the Forecast hub (dataframe)
  • download_variant_introduction.R
    • download_variant_introduction() - gets variants of concern dataset from ECDC; for each variant/country finds the first period with variant % of cases greater than 5% and up to 50%, before the first peak for that variant (dataframe)

• Functions to access evaluation

  • download_latest_eval.R
    • download_latest_eval() - downloads latest evaluation from the European Forecast Hub github repository (read more about Hub evaluation), and lightly cleans. (dataframe)

summarise

These scripts create numbers, figures, and tables used in the text and SI of the analysis.