Skip to contents

Calculate Negative Log-Likelihood for SIR Model

mll_sir.Rd

Computes the negative log-likelihood for the SIR model under the specified distributional family. Diagonal entries (self-ties) and NA values are excluded. Used as the objective function for parameter estimation.

Usage

mll_sir(tab, Y, W, X, Z, family, fix_receiver = FALSE)

Arguments

tab

Numeric vector of parameters [theta, alpha_2:p, beta_1:p].

Y

Three-dimensional array (m x m x T) of observed outcomes. Can contain NA values which are automatically excluded.

W

Three-dimensional array (m x m x p) of influence covariates, or NULL for no network influence.

X

Three-dimensional array (m x m x T) carrying network influence.

Z

Array of exogenous covariates (3D or 4D), or NULL.

family

Character string specifying the distribution:

  • "poisson": Count data with log link

  • "normal": Continuous data with identity link (assumes sigma=1)

  • "binomial": Binary data with logit link

fix_receiver

Logical. If TRUE, B is fixed to identity and tab is parsed as [theta, alpha_1:p]. Default FALSE.

Value

Numeric scalar giving the negative log-likelihood. Lower values indicate better fit. Used for optimization.

Note

The normal family assumes unit variance (sigma=1) for simplicity. The actual variance is estimated separately if needed.

Examples

if (FALSE) { # \dontrun{
# Poisson example
Y <- array(rpois(1000, 2), dim=c(10,10,10))
nll <- mll_sir(tab, Y, W, X, Z, "poisson")

# Binomial example with missing data
Y_binary <- array(rbinom(1000, 1, 0.3), dim=c(10,10,10))
Y_binary[1,1,1] <- NA  # Missing value
nll <- mll_sir(tab, Y_binary, W, X, Z, "binomial")
} # }