# standard_error: Component-wise standard errors of B In PLN-team/PLNmodels: Poisson Lognormal Models

 standard_error.PLNPCAfit R Documentation

## Component-wise standard errors of B

### Description

Extracts univariate standard errors for the estimated coefficient of B. Standard errors are computed from the (approximate) Fisher information matrix.

### Usage

``````## S3 method for class 'PLNPCAfit'
standard_error(
object,
type = c("variational", "jackknife", "sandwich"),
parameter = c("B", "Omega")
)

standard_error(
object,
type = c("variational", "jackknife", "sandwich"),
parameter = c("B", "Omega")
)

## S3 method for class 'PLNfit'
standard_error(
object,
type = c("variational", "jackknife", "bootstrap", "sandwich"),
parameter = c("B", "Omega")
)

## S3 method for class 'PLNfit_fixedcov'
standard_error(
object,
type = c("variational", "jackknife", "bootstrap", "sandwich"),
parameter = c("B", "Omega")
)

## S3 method for class 'PLNmixturefit'
standard_error(
object,
type = c("variational", "jackknife", "sandwich"),
parameter = c("B", "Omega")
)

## S3 method for class 'PLNnetworkfit'
standard_error(
object,
type = c("variational", "jackknife", "sandwich"),
parameter = c("B", "Omega")
)
``````

### Arguments

 `object` an R6 object with class PLNfit `type` string describing the type of variance approximation: "variational", "jackknife", "sandwich" (only for fixed covariance). Default is "variational". `parameter` string describing the target parameter: either B (regression coefficients) or Omega (inverse residual covariance)

### Value

A p * d positive matrix (same size as `B`) with standard errors for the coefficients of `B`

### Methods (by class)

• `standard_error(PLNPCAfit)`: Component-wise standard errors of B in `PLNPCAfit` (not implemented yet)

• `standard_error(PLNfit)`: Component-wise standard errors of B in `PLNfit`

• `standard_error(PLNfit_fixedcov)`: Component-wise standard errors of B in `PLNfit_fixedcov`

• `standard_error(PLNmixturefit)`: Component-wise standard errors of B in `PLNmixturefit` (not implemented yet)

• `standard_error(PLNnetworkfit)`: Component-wise standard errors of B in `PLNnetworkfit` (not implemented yet)

`vcov.PLNfit()` for the complete variance covariance estimation of the coefficient

### Examples

``````data(trichoptera)
trichoptera <- prepare_data(trichoptera\$Abundance, trichoptera\$Covariate)
myPLN <- PLN(Abundance ~ 1 + offset(log(Offset)), data = trichoptera,
control = PLN_param(config_post = list(variational_var = TRUE)))
standard_error(myPLN)
``````

PLN-team/PLNmodels documentation built on Sept. 11, 2024, 4:08 p.m.