Learn R Programming

gctsc (version 0.1.3)

predict.gctsc: Predictive Distribution and Scoring for Gaussian Copula Time Series Models

Description

Computes the one-step-ahead predictive distribution for a fitted Gaussian copula time series model, including summary statistics (mean, median, mode, variance) and optional scoring rules (CRPS and LOGS) if an observed value is supplied.

Usage

# S3 method for gctsc
predict(object, ..., method = "GHK",
                        y_obs = NULL, X_test = NULL)

Value

mean

Predictive mean of \(Y_{t+1}\).

median

Predictive median of \(Y_{t+1}\).

mode

Predictive mode of \(Y_{t+1}\).

variance

Predictive variance of \(Y_{t+1}\).

CRPS

Continuous Ranked Probability Score (if y_obs is provided).

LOGS

Logarithmic score (if y_obs is provided).

p_y

Predictive pmf over \((0,\dots,y_{\max})\).

lower, upper

95\(\%\) predictive interval bounds.

Arguments

object

A fitted model object of class gctsc.

...

Not used. Included for S3 method consistency.

method

Character string specifying the prediction method: "TMET" or "GHK" (default: "GHK").

y_obs

Optional observed value used to compute CRPS and LOGS.

X_test

Optional covariate information for prediction. Can be a named numeric vector, or a 1-row matrix/data.frame with column names matching the model covariates.

See Also

gctsc, arma.cormat