---
title: "8. Rolling networks"
output:
  rmarkdown::html_vignette:
    toc: true
vignette: >
  %\VignetteIndexEntry{8. Rolling networks}
  %\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)
```

Rolling windows ask whether a person's dynamics are **stable over time**. Each
window refits the estimator on a moving slice of the series.

# Rolling OLS VAR

Set `keep_fits = TRUE` to retain the per-window model objects for inspection and
plotting.

```{r rolling-var}
rolling_ols <- rolling_var(srl, vars = vars, id = "name", subject = "Grace",
                           window_size = 50, step = 20, scale = TRUE,
                           keep_fits = TRUE)

rolling_ols
```

The result is a tidy table by default — one row per window-by-edge estimate.
Use `head()` to peek; never index the table with brackets.

```{r rolling-var-table}
head(as.data.frame(rolling_ols))
```

Inspect a single window's matrices or plot it by passing `fit` — by name or by
index — instead of digging into `$fits`.

```{r rolling-var-matrices}
matrices(rolling_ols, fit = 1)
```

```{r plot-rolling-var, eval=has_cograph}
plot(rolling_ols, fit = 1, layer = "temporal")
```

# Rolling graphical VAR

`rolling_graphical_var()` does the same with sparse estimation.

```{r rolling-gvar}
rolling_gvar <- rolling_graphical_var(srl, vars = vars, id = "name",
                                      subject = "Grace", window_size = 70,
                                      step = 25, n_lambda = 8, gamma = 0,
                                      keep_fits = TRUE)

rolling_gvar

head(as.data.frame(rolling_gvar))
```

```{r plot-rolling-gvar, eval=has_cograph}
plot(rolling_gvar, fit = 1, layer = "temporal")
```