Getting Started with dcvar

The hardware and bandwidth for this mirror is donated by dogado GmbH, the Webhosting and Full Service-Cloud Provider. Check out our Wordpress Tutorial.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]dogado.de.

Installation

The dcvar package uses rstan as its default backend for Bayesian model fitting.

# Install dcvar from CRAN
install.packages("dcvar")

For the development version:

install.packages("remotes")
remotes::install_github("benlug/dcvar")

Optionally, you can use cmdstanr as an alternative backend by installing it and CmdStan:

install.packages("cmdstanr",
  repos = c("https://stan-dev.r-universe.dev", getOption("repos"))
)
cmdstanr::install_cmdstan()

Simulating Data

dcvar includes trajectory generators and a data simulator for testing:

library(dcvar)

# Generate a decreasing rho trajectory (therapy effect)
rho_true <- rho_decreasing(n_time = 150, rho_start = 0.7, rho_end = 0.3)

# Simulate bivariate VAR(1) data with this trajectory
sim <- simulate_dcvar(
  n_time = 150,
  rho_trajectory = rho_true,
  seed = 42
)

head(sim$Y_df)
#>   time        y1        y2
#> 1    1 0.0000000 0.0000000
#> 2    2 1.3709584 0.5378943
#> 3    3 0.8282054 1.0071266
#> 4    4 0.7534426 0.5867669
#> 5    5 1.7962315 1.2239371
#> 6    6 2.6796869 1.8905391

Fitting the DC-VAR Model

The main fitting function is dcvar(). It accepts a data frame and the names of two variables to model:

fit <- dcvar(
  data = sim$Y_df,
  vars = c("y1", "y2"),
  chains = 4,
  iter_warmup = 1000,
  iter_sampling = 2000,
  seed = 123
)

Inspecting Results

# Quick overview
print(fit)

# Detailed summary
summary(fit)

# Posterior means for VAR parameters
coef(fit)

# Time-varying rho as a data frame
rho_df <- rho_trajectory(fit)
head(rho_df)

# MCMC diagnostics
dcvar_diagnostics(fit)

Fitted Values and Predictions

The fitted() method returns one-step-ahead fitted values from the VAR(1) component. The predict() method adds marginal prediction intervals for normal margin fits:

# VAR(1) fitted values: y_hat[t] = mu + Phi * (y[t-1] - mu)
fit_df <- fitted(fit)
head(fit_df)

# Predictions with 95% marginal intervals
pred_df <- predict(fit)
head(pred_df)

# Custom interval level
pred_90 <- predict(fit, ci_level = 0.90)

Tidy Parameter Summary

Use as.data.frame() to export the full parameter summary as a tidy data frame:

param_df <- as.data.frame(fit)
head(param_df)

Clinical Interpretation

interpret_rho_trajectory() provides a human-readable summary of the estimated dependence trajectory:

interpret_rho_trajectory(fit)

Visualisation

# Rho trajectory with credible intervals
# Red line = true trajectory (simulation only)
plot_rho(fit, true_rho = sim$true_params$rho)

# VAR coefficient heatmap
plot_phi(fit)

# MCMC diagnostics (trace, Rhat, ESS)
plot(fit, type = "diagnostics")

# Posterior predictive check
# Currently available for normal and exponential margins
plot_ppc(fit)

LOO-CV Model Comparison

Compare the DC-VAR against a constant-correlation baseline:

fit_constant <- dcvar_constant(sim$Y_df, vars = c("y1", "y2"))

dcvar_compare(dcvar = fit, constant = fit_constant)

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.
Health stats visible at Monitor.