Methods for local and federated BFI objects

Print, coefficient and variance methods for objects returned by MAP.estimation and bfi.

Usage

# S3 method for class 'bfi'
coef(object, ...)

# S3 method for class 'bfi'
print(x, digits = max(3L, getOption("digits") - 3L), ...)

# S3 method for class 'bfi'
vcov(object, ...)

Arguments

...

Further arguments (currently ignored).

x, object

An object inheriting from "bfi", including "bfi_local" and "bfi_federated" objects.

digits

Number of significant digits used for printing.

Value

print() returns x invisibly. coef() returns a named numeric vector of the MAP or BFI estimates. vcov() returns the approximate posterior covariance matrix.

Details

vcov() returns the approximate posterior covariance matrix corresponding to the parameterization returned by coef(). For Gaussian models, the inverse curvature matrix is defined on the \(\log(\sigma^2)\) scale for residual-variance parameters; these rows and columns are transformed to the original \(\sigma^2\) scale using the multivariate delta method.

For weighted models, the returned covariance matrix is a curvature-based approximation conditional on the weights used in estimation. It does not automatically account for uncertainty associated with estimated propensity scores.

See also

summary.bfi

Examples

X <- data.frame(x1 = rnorm(50))
y <- rnorm(50)
Lambda <- inv.prior.cov(X, lambda = 0.01, family = "gaussian")
fit <- MAP.estimation(y, X, family = "gaussian", Lambda = Lambda)
fit
#> 
#> Local MAP estimates (family: gaussian)
#> 
#> (Intercept)           x1       sigma2  
#>   -0.129876     0.003659     0.872124  
#> 
coef(fit)
#>  (Intercept)           x1       sigma2 
#> -0.129875687  0.003659374  0.872123970 
vcov(fit)
#>               (Intercept)            x1        sigma2
#> (Intercept)  1.744312e-02 -2.688538e-04  7.966818e-07
#> x1          -2.688538e-04  1.963049e-02 -3.800053e-08
#> sigma2       7.966818e-07 -3.800053e-08  3.041351e-02