Tidy hierarchical forecast aggregation results — tidy_forecast

Converts the stage tables returned by aggregate_model_forecasts() into a single long table. The result is useful for diagnostics and plotting, and retains the identifiers that apply at each aggregation level.

Usage

tidy_forecast_aggregation(
  aggregation,
  unit_by,
  outcomes = NULL,
  completion_col = "completion",
  prompt_col = "prompt_id",
  model_col = "model_id",
  family_col = "family"
)

Arguments

aggregation: A named list returned by aggregate_model_forecasts().
unit_by: Character vector of columns identifying a forecast target.
outcomes: Optional character vector of numeric outcome columns. When omitted, numeric columns other than count and identifier columns are used.
completion_col, prompt_col, model_col, family_col: Names of the identifier columns used when the aggregation was made. family_col may be NULL.

Value

A tibble with aggregation_level, the available identifiers, outcome, estimate, contributor_type, and n_contributors.