Compute resolution of forecasts, normalised by the uncertainty

Source: R/misc.R

The resolution is computed as the mean squared distance to a base rate (reference forecast) and is then normalised by the uncertainty (maximum resolution). This means the output is between 0 and 1, 1 corresponding to the maximum resolution.

compute_resolution(f, p0)

Arguments

f

Vector of forecasts

p0

Vector of base rate. In the case rate is usually the prevalence of a uniform forecast (e.g. 1 / number of categories) but can depend on the observation (hence the vector).

Value

Vector of resolution values

Examples

compute_resolution(seq(0, 1, .1), 0.5)
#> [1] 0.4