Create a bi-variate histogram of forecast, observation pairs

Values of forecasts and observations are binned into bands whereby the density of forecast, observation pairs for each bin is calculated. Under the hood, the data are binned into hexagons using hexbin. Hexagons are used since they have symmetry with nearest neighbours unlike square bins, and at plot time they are the polygon with the maximum number of sides that tessellate.

Usage

bin_fcst_obs(
  .fcst,
  parameter,
  groupings = "lead_time",
  num_bins = 30,
  show_progress = TRUE,
  ...
)

# S3 method for harp_det_point_df
bin_fcst_obs(
  .fcst,
  parameter,
  groupings = "lead_time",
  num_bins = 30,
  show_progress = TRUE,
  fcst_model = NULL,
  ...
)

# S3 method for harp_ens_point_df
bin_fcst_obs(
  .fcst,
  parameter,
  groupings = "lead_time",
  num_bins = 30,
  show_progress = TRUE,
  fcst_model = NULL,
  ...
)

Arguments

.fcst: A harp_df data frame or a harp_list.
parameter: The column containing the parameter. Can be unquoted, a quoted string, or an embraced variable name (i.e var).
groupings: The groupings for which to compute the binned densities. Must be a vector of strings, or a list of vectors of strings.
num_bins: The number of bins into which to partition the observations.
show_progress: Logical. Whether to show a progress bar.
...: Arguments for methods.
fcst_model: The name of the forecast model. If .fcst does not contain a fcst_model, a new column is created and populated with this value. If a fcst_model column exists, the value in the column is replaced with this value.

Value

A harp_verif list.