predict.singleRStaticCountData {singleRcapture} | R Documentation |
Predict method for singleRStaticCountData
class
Description
A method for predict
function, works analogous to predict.glm
but gives the possibility to get standard errors of
mean/distribution parameters and directly get pop size estimates for new data.
Usage
## S3 method for class 'singleRStaticCountData'
predict(
object,
newdata,
type = c("response", "link", "mean", "popSize", "contr"),
se.fit = FALSE,
na.action = NULL,
weights,
cov,
...
)
Arguments
object |
an object of |
newdata |
an optional |
type |
the type of prediction required, possible values are:
by default set to |
se.fit |
a logical value indicating whether standard errors should be
computed. Only matters for |
na.action |
does nothing yet. |
weights |
optional vector of weights for |
cov |
optional matrix or function or character specifying either
a covariance matrix or a function to compute that covariance matrix.
By default |
... |
arguments passed to other functions, for now this only affects
|
Details
Standard errors are computed with assumption of regression coefficients being asymptotically normally distributed, if this assumption holds then each of linear predictors i.e. each row of \(\boldsymbol{\eta}=\boldsymbol{X}_{vlm}\boldsymbol{\beta}\) is asymptotically normally distributed and their variances are expressed by well known formula. The mean \(\mu\) and distribution parameters are then differentiable functions of asymptotically normally distributed variables and therefore their variances can be computed using (multivariate) delta method.
Value
Depending on type
argument if one of "response", "link", "mean"
a matrix with fitted values and possibly standard errors if se.fit
argument was set to TRUE
, if type
was set to "contr"
a vector with inverses of probabilities, finally for "popSize"
an object of class popSizeEstResults
with its own methods containing
population size estimation results.
See Also
redoPopEstimation()
stats::summary.glm()
estimatePopsize()