Apply fitted precision weights to new observations

Uses the estimated phi and beta1 from a precision_weights fit to compute \(\sigma = \phi\, s^{\beta_1}\) and \(w = 1/\sigma^2\) for new observations — typically a precision grid (to draw a continuous weight profile) or held-out samples — without refitting.

Usage

predict_weights(object, newdata = NULL, ...)

Arguments

object

A precision_weights object from fit_precision_weights().

newdata

One of: a weight_data frame, a plain data frame with the predictor column, or a calibration_result(_multiplate) (which is run through as_weight_data(source = "grid") using the fit's design). NULL (default) re-applies to the fit's own data via the stored weights.

...

Passed to as_weight_data() when newdata is an ecosystem object.

Value

A data frame with the predictor, sigma, w, and w_norm columns added.

See also

fit_precision_weights()