Predict from an uncounted model

Returns predicted values from a fitted uncounted model, optionally for new data. This method enables compatibility with the marginaleffects package.

Usage

# S3 method for class 'uncounted'
predict(object, newdata = NULL, type = c("response", "link"), ...)

Arguments

object

An uncounted object.

newdata

Optional data frame for predictions. If NULL, returns fitted values from the original data.

type

Character: "response" (default) returns predicted counts \(\hat{\mu}_i = \exp(\mathbf{z}_i'\hat{\boldsymbol{\theta}})\); "link" returns the linear predictor \(\mathbf{z}_i'\hat{\boldsymbol{\theta}}\).

...

Additional arguments (ignored).

Value

A numeric vector of predicted values.

Examples

data(irregular_migration)
d <- irregular_migration[irregular_migration$year == "2019", ]
fit <- estimate_hidden_pop(d, ~ m, ~ n, ~ N, method = "poisson",
                           gamma = 0.005)
head(predict(fit))
#> [1]  2.5691032  0.2256555  2.1005157  1.9776050 22.8677881  0.5587853
head(predict(fit, newdata = d[1:5, ]))
#> [1]  2.5691032  0.2256555  2.1005157  1.9776050 22.8677881
head(predict(fit, type = "link"))
#> [1]  0.9435569 -1.4887458  0.7421829  0.6818865  3.1297293 -0.5819900