Skip to contents

Loss Function for Fitting the DKP Model

Source: R/loss_fun_dkp.R
loss_fun_dkp.Rd

Computes the loss used to fit the DKP model. Supports the Brier score (mean squared error) and negative log-loss (cross-entropy), under different prior specifications.

Usage

loss_fun_dkp(
  gamma,
  Xnorm,
  Y,
  prior = c("noninformative", "fixed", "adaptive"),
  r0 = 2,
  p0 = NULL,
  loss = c("brier", "log_loss"),
  kernel = c("gaussian", "matern52", "matern32")
)

Arguments

gamma

A numeric vector of log-transformed kernel hyperparameters.

Xnorm

A numeric matrix of normalized inputs (each column scaled to [0,1]).

Y

Matrix of observed multinomial counts, with dimension \(n \times q\).

prior

Type of prior to use. One of "noninformative", "fixed", or "adaptive".

r0

Global prior precision (only used when prior = "fixed" or "adaptive").

p0

Global prior mean vector (only used when prior = "fixed"). Must be of length \(q\).

loss

Loss function for kernel hyperparameter tuning. One of "brier" (default) or "log_loss".

kernel

Kernel function for local weighting. Choose from "gaussian", "matern52", or "matern32".

Value

A single numeric value representing the total loss (to be minimized).

See also

loss_fun, fit.DKP, get_prior_dkp, kernel_matrix

Examples

set.seed(123)
n = 10
Xnorm = matrix(runif(2 * n), ncol = 2)
m = rep(10, n)
y = rbinom(n, size = m, prob = runif(n))
Y = cbind(y, m-y)
loss_fun_dkp(gamma = rep(0, 2), Xnorm = Xnorm, Y = Y)
#> [1] 0.01862653