---
title: "3. Graphical VAR"
output:
  rmarkdown::html_vignette:
    toc: true
vignette: >
  %\VignetteIndexEntry{3. Graphical VAR}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(
  collapse = TRUE, comment = "#>", message = FALSE, warning = FALSE,
  fig.width = 7, fig.height = 5.5
)
```

```{r libraries}
library(idiographic)
data(srl)
vars <- c("efficacy", "value", "planning", "monitoring", "effort")
has_cograph <- requireNamespace("cograph", quietly = TRUE)
```

`graphical_var()` estimates **sparse** temporal and contemporaneous networks
with lasso regularization and EBIC model selection. It is the closest analogue
to the chapter's graphical VAR section, and the regularized counterpart to the
OLS baseline in the *Ordinary VAR* vignette.

# Fit one person

```{r gvar}
gvar_fit <- graphical_var(srl, vars = vars, id = "name", subject = "Grace",
                          n_lambda = 12, gamma = 0)

gvar_fit
```

The printout reports the selected EBIC, `gamma`, and the two tuning parameters,
then shows the sparse temporal and contemporaneous networks. Many cells are
exactly zero — that sparsity is the point of the method.

# Tidy tables

```{r gvar-tables}
head(edges(gvar_fit))

nodes(gvar_fit)

summary(gvar_fit)
```

```{r gvar-matrices}
matrices(gvar_fit)
```

# Plot

```{r plot-gvar, eval=has_cograph}
plot(gvar_fit)
```

```{r plot-gvar-temporal, eval=has_cograph}
plot(gvar_fit, layer = "temporal")
```

```{r plot-gvar-contemp, eval=has_cograph}
plot(gvar_fit, layer = "contemporaneous")
```