Package 'idiographic'

Description

Returns netobjects unchanged; promotes a bare cograph_network.

Usage

as_netobject(x, ...)

Arguments

Value

A c("netobject", "cograph_network") object.

Description

Returns the two networks a graphical VAR contains as Nestimate netobjects, so each renders directly with cograph::splot() (or any netobject verb) without the caller transposing matrices or dropping intercept columns. The temporal network is oriented [from = predictor(t-1), to = outcome(t)].

Usage

## S3 method for class 'gvar_result'
as_netobject(x, ...)

Arguments

Value

A netobject_group: a named list with $temporal (directed) and $contemporaneous (undirected) netobjects.

Description

Returns the GIMME result as matrix-backed netobjects. By default these encode the same quantity the gimme package plots — the proportion of subjects that have each path (path_counts / n_subjects) — not the group-average coefficient (which dilutes toward zero and is not what GIMME displays). For the faithful single mixed network (dashed lag / solid contemporaneous, group/individual colouring, autoregressive self-loops) use plot_gimme().

Usage

## S3 method for class 'net_gimme'
as_netobject(x, style = c("pnode", "unified"), weight = c("prop", "coef"), ...)

Arguments

Value

For style = "pnode", a netobject_group with ⁠$temporal⁠ and ⁠$contemporaneous⁠. For style = "unified", one c("netobject", "cograph_network") object with ⁠2p⁠ nodes.

See Also

plot_gimme() for the faithful gimme-style mixed plot.

Description

Returns the three networks as netobjects oriented for plotting (temporal edges run predictor -> outcome, matching graphical_var), so cograph::splot() renders them consistently. The raw fit$temporal$weights keep mlVAR's ⁠[outcome, predictor]⁠ layout for equivalence.

Usage

## S3 method for class 'net_mlvar'
as_netobject(x, ...)

Arguments

Value

A netobject_group with ⁠$temporal⁠, ⁠$contemporaneous⁠, ⁠$between⁠.

Description

Coerce a var_bayes_result to plottable netobjects

Usage

## S3 method for class 'var_bayes_result'
as_netobject(x, ...)

Arguments

Value

A netobject_group with temporal (directed) and contemporaneous (undirected) netobjects.

Description

Builds the same lag-1 design used by graphical_var() and build_var(), then returns tidy diagnostics for missingness, day-boundary drops, simple linear trends, AR(1) persistence, split-half mean/variance drift, an ADF-style unit-root screen, and zero-variance variables. This is a preflight tool: it does not fit a network, but it makes the modelling input explicit before estimating VAR, graphical VAR, uSEM, GIMME, or mlVAR models.

Usage

audit_preprocess(
  data,
  vars,
  id = NULL,
  day = NULL,
  beep = NULL,
  scale = TRUE,
  center_within = TRUE,
  delete_missings = TRUE,
  min_obs = NULL,
  subject = NULL,
  trend_alpha = 0.05,
  ar_threshold = 0.95,
  mean_shift_threshold = 0.8,
  sd_ratio_threshold = 2,
  unit_root_t_cutoff = -2.86
)

Arguments

Value

A preprocess_audit object with pairs, counts, diagnostics, and matrices.

See Also

build_var(), graphical_var()

Examples

set.seed(1)
d <- data.frame(id = 1, day = 1, beep = 1:40,
                A = rnorm(40), B = rnorm(40))
audit <- audit_preprocess(d, vars = c("A", "B"), id = "id",
                          day = "day", beep = "beep")
audit$counts
audit$diagnostics

Description

Estimates person-specific directed networks from intensive longitudinal data using the unified Structural Equation Modeling (uSEM) framework. Implements a data-driven search that identifies:

1. Group-level paths: Directed edges present for a majority (default 75\

2. Individual-level paths: Additional edges specific to each person, found after group paths are established.

Uses lavaan for SEM estimation and modification indices. Accepts a single data frame with an ID column (not CSV directories).

Usage

build_gimme(
  data,
  vars,
  id,
  time = NULL,
  day = NULL,
  beep = NULL,
  min_obs = NULL,
  subject = NULL,
  ar = TRUE,
  standardize = FALSE,
  groupcutoff = 0.75,
  subcutoff = 0.75,
  paths = NULL,
  exogenous = NULL,
  hybrid = FALSE,
  VAR = FALSE,
  rmsea_cutoff = 0.05,
  srmr_cutoff = 0.05,
  nnfi_cutoff = 0.95,
  cfi_cutoff = 0.95,
  n_excellent = 2L,
  seed = NULL,
  group_correct = "Bonferoni Group",
  subgroup = FALSE,
  outcome = NULL,
  conv_vars = NULL,
  mult_vars = NULL,
  lv_model = NULL,
  lasso_model_crit = NULL,
  ms_allow = FALSE,
  ordered = NULL,
  dir_prop_cutoff = 0,
  out = NULL,
  sep = NULL,
  header = NULL,
  plot = FALSE,
  sub_feature = "lag & contemp",
  sub_method = "Walktrap",
  sub_sim_thresh = "lowest",
  confirm_subgroup = NULL,
  conv_length = 16,
  conv_interval = 1,
  mean_center_mult = FALSE,
  diagnos = FALSE,
  ms_tol = 1e-05,
  lv_estimator = "miiv",
  lv_scores = "regression",
  lv_miiv_scaling = "first.indicator",
  lv_final_estimator = "miiv"
)

Arguments

Value

An S3 object of class "net_gimme" containing:

temporal

p x p matrix of group-level temporal (lagged) path counts – entry [i,j] = number of individuals with path j(t-1)->i(t).

contemporaneous

p x p matrix of group-level contemporaneous path counts – entry [i,j] = number of individuals with path j(t)->i(t).

coefs

List of per-person p x 2p coefficient matrices (rows = endogenous, cols = [lagged, contemporaneous]).

psi

List of per-person 2p x 2p residual covariance matrices over c(lag_names, varnames) (the standardized lavaan psi block), not the p x p current-variable block alone.

fit

Data frame of per-person fit indices (chisq, df, pvalue, rmsea, srmr, nnfi, cfi, bic, aic, logl, status).

path_counts

p x 2p matrix: how many individuals have each path.

paths

List of per-person character vectors of lavaan path syntax.

group_paths

Character vector of group-level paths found.

individual_paths

List of per-person character vectors of individual-level paths (beyond group).

syntax

List of per-person full lavaan syntax strings.

labels

Character vector of variable names.

n_subjects

Integer. Number of individuals.

n_obs

Integer vector. Time points per individual.

config

List of configuration parameters.

See Also

build_mlvar, graphical_var, as_netobject

Examples

# Create simple panel data (3 subjects, 4 variables, 50 time points).
set.seed(42)
n_sub <- 3; n_t <- 50; vars <- paste0("V", 1:4)
rows <- lapply(seq_len(n_sub), function(i) {
  d <- as.data.frame(matrix(rnorm(n_t * 4), ncol = 4))
  names(d) <- vars; d$id <- i; d
})
panel <- do.call(rbind, rows)
res <- build_gimme(panel, vars = vars, id = "id")
print(res)

Description

Estimates three networks from ESM/EMA panel data, matching mlVAR::mlVAR() with estimator = "lmer", temporal = "fixed", contemporaneous = "fixed" at machine precision: (1) a directed temporal network of fixed-effect lagged regression coefficients, (2) an undirected contemporaneous network of partial correlations among residuals, and (3) an undirected between-subjects network of partial correlations derived from the person-mean fixed effects.

Usage

build_mlvar(
  data,
  vars,
  id,
  day = NULL,
  beep = NULL,
  lags = 1L,
  estimator = c("lmer", "default", "lm", "Mplus"),
  temporal = c("fixed", "correlated", "orthogonal", "unique", "default"),
  contemporaneous = c("fixed", "correlated", "orthogonal", "unique", "default"),
  AR = FALSE,
  scale = FALSE,
  scaleWithin = FALSE,
  nCores = 1L,
  verbose = FALSE,
  lag = NULL,
  standardize = NULL,
  min_obs = NULL,
  subject = NULL
)

Arguments

Details

The algorithm follows mlVAR's lmer pipeline exactly:

1. Drop rows with NA in id/day/beep and optionally grand-mean standardize each variable.

2. Expand the per-(id, day) beep grid and right-join original values, producing the augmented panel (augData).

3. Add within-person lagged predictors (⁠L1_*⁠) and person-mean predictors (⁠PM_*⁠).

4. For each outcome variable fit lmer(y ~ within + between-except-own-PM + (1 | id)) with REML = FALSE. Collect the fixed-effect temporal matrix B, between-effect matrix Gamma, random-intercept SDs (mu_SD), and lmer residual SDs.

5. Contemporaneous network: cor2pcor(D %*% cov2cor(cor(resid)) %*% D).

6. Between-subjects network: cor2pcor(pseudoinverse(forcePositive(D (I - Gamma)))).

Validated to machine precision (max_diff < 1e-10) against mlVAR::mlVAR() on 25 real ESM datasets from openesm and 20 simulated configurations (seeds 201-220). See tmp/mlvar_equivalence_real20.R and tmp/mlvar_equivalence_20seeds.R.

Value

A dual-class c("net_mlvar", "netobject_group") object — a named list of three full netobjects, one per network, plus model-level metadata stored as attributes. Each element is a standard c("netobject", "cograph_network") weight-matrix wrapper (no raw ⁠$data⁠), so print(), summary(), coefs(), and cograph::splot(fit$temporal) work directly. The three constituents are matrix-wrapped and carry no underlying panel data, so any data-resampling workflow (bootstrap, reliability, stability) must start from the original panel rather than from these wrappers. Structure:

fit$temporal

Directed netobject for the ⁠d x d⁠ matrix of fixed-effect lagged coefficients. ⁠$weights[i, j]⁠ is the effect of variable j at t-lag on variable i at t. method = "mlvar_temporal", directed = TRUE.

fit$contemporaneous

Undirected netobject for the ⁠d x d⁠ partial-correlation network of within-person lmer residuals. method = "mlvar_contemporaneous", directed = FALSE.

fit$between

Undirected netobject for the ⁠d x d⁠ partial-correlation network of person means, derived from D (I - Gamma). method = "mlvar_between", directed = FALSE. Convention: when a random-intercept SD is 0 the between network is not estimable; idiographic returns an all-zero matrix (with a warning) as a plotting-oriented convention, whereas mlVAR returns an all-NA matrix. The contemporaneous network follows the same zero-on-degeneracy convention. This is a deliberate departure from strict reference equivalence in the singular case.

attr(fit, "coefs") / coefs()

Tidy data.frame with one row per ⁠(outcome, predictor)⁠ pair and columns outcome, predictor, beta, se, t, p, ci_lower, ci_upper, significant. Filter, sort, or plot with base R or the tidyverse. Retrieve with coefs(fit).

attr(fit, "n_obs")

Number of rows in the augmented panel after na.omit.

attr(fit, "n_subjects")

Number of unique subjects remaining.

attr(fit, "lag")

Lag order used.

attr(fit, "standardize")

Logical; whether pre-augmentation standardization was applied.

See Also

build_gimme(), graphical_var(), as_netobject()

Examples

set.seed(1)
n_id <- 8; n_t <- 30; vars <- c("A", "B", "C")
rows <- lapply(seq_len(n_id), function(i) {
  m <- as.data.frame(matrix(rnorm(n_t * 3), ncol = 3))
  names(m) <- vars
  m$id <- i; m$day <- 1L; m$beep <- seq_len(n_t)
  m
})
d <- do.call(rbind, rows)
fit <- build_mlvar(d, vars = vars, id = "id", day = "day", beep = "beep")
print(fit)
summary(fit)

Description

Native, pure-R Bayesian estimator for a two-level VAR(1) that statistically reproduces Mplus DSEM output (the estimator behind mlVAR::mlVAR(estimator = "Mplus")) without needing Mplus installed. A conjugate Gibbs sampler estimates a fixed temporal matrix, a within-person residual (contemporaneous) network, and a between-person network, using latent mean centering and Mplus's default priors. Point estimates are posterior medians with posterior SDs and 95% credible intervals.

Usage

build_mlvar_bayes(
  data,
  vars,
  id,
  day = NULL,
  beep = NULL,
  lags = 1L,
  temporal = c("fixed", "default", "random"),
  contemporaneous = c("fixed", "default"),
  residual = c("fixed", "random"),
  scale = TRUE,
  scaleWithin = FALSE,
  tinterval = NULL,
  impute = FALSE,
  n_iter = 4000L,
  n_burnin = NULL,
  n_chains = 2L,
  thin = 1L,
  seed = NULL,
  min_obs = NULL,
  subject = NULL,
  verbose = FALSE
)

Arguments

Details

The sampler alternates five conjugate full-conditional draws per iteration: the latent person means mu_i (Gaussian), the fixed temporal matrix B (matrix-normal), the within residual covariance Sigma_W (inverse-Wishart), the grand mean alpha (Gaussian), and the between covariance Sigma_B (inverse-Wishart). The lagged predictor is recentred on the current mu_i draw every iteration (latent mean centering). Data are globally standardized first (matching mlVAR's scale = TRUE); the first observation of each block is used only as a lag (condition-on-first).

Validated to statistical (Monte-Carlo-error) equivalence against real Mplus 9 DSEM output on standardized synthetic panels: posterior medians of B, Sigma_W, Sigma_B agree with Mplus to well within a posterior SD.

Value

A net_mlvar_bayes object (also inheriting net_mlvar), a named list of three netobjects (temporal, contemporaneous, between) with posterior-summary attributes. coefs() returns a tidy table with estimate (posterior median), posterior_sd, ci_lower, ci_upper, p (one-tailed), and significant (95% CI excludes 0). Posterior draws and the max Gelman-Rubin PSR are kept in attributes.

See Also

build_mlvar() (frequentist lmer path), build_mlvar_mplus() (true-Mplus wrapper).

Examples

set.seed(1)
n_id <- 10; n_t <- 40; vars <- c("A", "B")
rows <- lapply(seq_len(n_id), function(i) {
  y <- matrix(0, n_t, 2)
  for (t in 2:n_t) y[t, ] <- c(0.3, 0.15) * y[t - 1, ] + rnorm(2)
  data.frame(id = i, beep = seq_len(n_t), A = y[, 1], B = y[, 2])
})
d <- do.call(rbind, rows)
fit <- build_mlvar_bayes(d, vars = vars, id = "id", beep = "beep",
                         n_iter = 2000, seed = 1)
print(fit)
coefs(fit)

Description

Runs the Mplus Bayesian estimator exposed by mlVAR::mlVAR(estimator = "Mplus") and converts the returned posterior summaries into idiographic's network/tidy accessors. This is a true Mplus backend: Mplus must be installed and discoverable by MplusAutomation::detectMplus().

Usage

build_mlvar_mplus(
  data,
  vars,
  id,
  day = NULL,
  beep = NULL,
  lags = 1L,
  temporal = c("fixed", "correlated", "orthogonal", "default"),
  contemporaneous = c("fixed", "correlated", "orthogonal", "default"),
  nCores = 1L,
  scale = TRUE,
  scaleWithin = FALSE,
  MplusSave = TRUE,
  MplusName = "mlVAR_mplus",
  iterations = "(2000)",
  chains = nCores,
  signs,
  min_obs = NULL,
  subject = NULL,
  workdir = NULL,
  verbose = TRUE,
  ...
)

Arguments

Value

A net_mplus object, also inheriting from net_mlvar, with temporal, contemporaneous, and between networks plus Mplus metadata in attributes. The original mlVAR/Mplus object is available as attr(x, "mplus").

See Also

build_mlvar()

Examples

## Not run: 
fit <- build_mlvar_mplus(
  data, vars = c("A", "B", "C"), id = "id", beep = "time",
  temporal = "fixed", contemporaneous = "fixed",
  MplusName = "my_mplus_mlvar"
)
edges(fit)
attr(fit, "mplus")$output$summaries

## End(Not run)

Description

Fits person-specific unified Structural Equation Models (uSEM) for intensive longitudinal data. With trim = FALSE, the model is fixed by temporal, contemporaneous, residual_cov, and paths. With trim = TRUE, idiographic uses an independent clean-room modification-index entry and z-value pruning layer over the declared candidate set.

Usage

build_usem(
  data,
  vars,
  id,
  time = NULL,
  day = NULL,
  beep = NULL,
  min_obs = NULL,
  subject = NULL,
  temporal = c("ar", "all", "none"),
  contemporaneous = c("none", "all"),
  residual_cov = TRUE,
  trim = FALSE,
  trim_alpha = 0.05,
  trim_fit_criteria = 3L,
  cfi_cutoff = 0.95,
  tli_cutoff = 0.95,
  rmsea_cutoff = 0.08,
  srmr_cutoff = 0.08,
  paths = NULL,
  exogenous = NULL,
  standardize = FALSE,
  estimator = "ml",
  seed = NULL
)

Arguments

Value

A net_usem object with average temporal, contemporaneous, and residual-covariance networks, per-subject matrices, fit indices, syntax, and tidy coefficients.

See Also

build_gimme(), graphical_var(), build_mlvar()

Description

Fits a transparent VAR(1) baseline from intensive longitudinal data using ordinary least squares. The lag construction, scaling, within-person centering, and day-boundary behavior match graphical_var(), but no regularization or EBIC model selection is applied.

Usage

build_var(
  data,
  vars,
  id = NULL,
  day = NULL,
  beep = NULL,
  lags = 1L,
  scale = TRUE,
  center_within = TRUE,
  delete_missings = TRUE,
  min_obs = NULL,
  subject = NULL
)

Arguments

Value

A var_result object with temporal OLS coefficients, residual covariance, residual precision, contemporaneous partial correlations, and tidy accessors.

See Also

graphical_var(), build_usem()

Description

Native, pure-R Bayesian VAR(1) that reproduces Mplus's Bayesian (DSEM/time-series) estimates without needing Mplus. It is the unregularized Bayesian counterpart of graphical_var(): instead of a graphical-lasso / EBIC sparse fit, it estimates a full VAR(1) with a flat prior on the temporal coefficients and an inverse-Wishart prior on the residual precision, then reports the temporal network B and the contemporaneous partial-correlation network derived from the residual covariance. With more than one subject the data are within-person centred and pooled (as in graphical_var()).

Usage

build_var_bayes(
  data,
  vars,
  id = NULL,
  day = NULL,
  beep = NULL,
  lags = 1L,
  scale = TRUE,
  center_within = TRUE,
  n_iter = 4000L,
  n_burnin = NULL,
  n_chains = 2L,
  thin = 1L,
  seed = NULL,
  min_obs = NULL,
  subject = NULL,
  verbose = FALSE
)

Arguments

Value

A var_bayes_result object (a cograph group with temporal and contemporaneous netobjects) carrying beta, temporal, kappa, PCC, PDC, posterior draws, and a tidy coefs() table (posterior median, SD, 95% CI, one-tailed p, significance by CI excluding 0).

See Also

graphical_var() (regularized GLASSO/EBIC), build_var() (OLS), build_mlvar_bayes() (multilevel Bayesian VAR).

Examples

set.seed(1)
y <- matrix(0, 200, 2)
for (t in 2:200) y[t, ] <- c(0.4, 0.3) * y[t - 1, ] + rnorm(2)
d <- data.frame(A = y[, 1], B = y[, 2])
fit <- build_var_bayes(d, vars = c("A", "B"), n_iter = 2000, seed = 1)
print(fit)
coefs(fit)

Description

Applies build_var() to each subject separately, returning one transparent person-specific OLS VAR result per individual. This is the unregularized companion to graphical_var_each() and is useful as an equivalence baseline for checking lag construction, scaling, and temporal coefficient direction.

Usage

build_var_each(
  data,
  vars,
  id,
  day = NULL,
  beep = NULL,
  min_obs = NULL,
  ...
)

Arguments

Value

A named list of var_result objects (class var_list), one element per subject, named by subject id. Subjects that cannot be fit are dropped with a warning.

See Also

build_var(), graphical_var_each()

Examples

set.seed(1)
d <- data.frame(
  id = rep(1:3, each = 40),
  day = rep(1, 120),
  beep = rep(seq_len(40), 3),
  A = rnorm(120), B = rnorm(120), C = rnorm(120)
)
fits <- build_var_each(d, vars = c("A", "B", "C"), id = "id",
                       day = "day", beep = "beep")
fits[["1"]]

Description

Generic accessor for the tidy coefficient table stored on a build_mlvar() result. Returns a data.frame with one row per ⁠(outcome, predictor)⁠ pair and columns outcome, predictor, beta, se, t, p, ci_lower, ci_upper, significant.

Usage

coefs(x, ...)

## S3 method for class 'net_mlvar'
coefs(x, ...)

## Default S3 method:
coefs(x, ...)

## S3 method for class 'net_mlvar_bayes'
coefs(x, ...)

## S3 method for class 'net_usem'
coefs(x, ...)

## S3 method for class 'var_result'
coefs(x, ...)

## S3 method for class 'var_bayes_result'
coefs(x, ...)

## S3 method for class 'gvar_result'
coefs(x, ...)

## S3 method for class 'net_gimme'
coefs(x, ...)

Arguments

Details

Only the within-person (temporal) coefficients are tabulated — these are the lagged fixed effects that populate fit$temporal. The between-subjects effects that go into fit$between are handled via the D (I - Gamma) transformation and are not exposed as a separate tidy table.

Value

A tidy data.frame of coefficient estimates.

Examples

set.seed(1)
n_id <- 8; n_t <- 30; vars <- c("A", "B", "C")
rows <- lapply(seq_len(n_id), function(i) {
  m <- as.data.frame(matrix(rnorm(n_t * 3), ncol = 3))
  names(m) <- vars
  m$id <- i; m$day <- 1L; m$beep <- seq_len(n_t)
  m
})
d <- do.call(rbind, rows)
fit <- build_mlvar(d, vars = vars, id = "id", day = "day", beep = "beep")
print(fit)
summary(fit)

Description

Fits one or more idiographic estimators to the same data and returns a tidy per-method/per-network comparison table. This is a reporting layer: it does not define a new model, and each row is computed from the estimator's own summary() method plus common edge-table accessors.

Usage

compare_idiographic(
  data,
  vars,
  estimators = c("var", "graphical_var"),
  id = NULL,
  day = NULL,
  beep = NULL,
  estimator_args = list(),
  keep_fits = FALSE
)

Arguments

Value

A model_comparison object with $comparison, $failures, and optionally $fits. $comparison is a tidy data.frame with one row per method/network.

See Also

build_var(), graphical_var(), estimate_stability()

Examples

set.seed(1)
d <- data.frame(id = 1, day = rep(1:4, each = 15),
                beep = rep(1:15, 4),
                A = rnorm(60), B = rnorm(60), C = rnorm(60))
cmp <- compare_idiographic(
  d, vars = c("A", "B", "C"), id = "id", day = "day", beep = "beep",
  estimators = c("var", "graphical_var"),
  estimator_args = list(graphical_var = list(n_lambda = 5))
)
cmp$comparison

Description

A single tidy verb for every network idiographic produces. Returns one row per edge with columns network (e.g. "temporal", "contemporaneous", "between"), from, to, weight – and, for GIMME, level ("group"/"individual"). Directed networks (temporal) keep every edge; undirected networks (contemporaneous, between) report each pair once.

Usage

## S3 method for class 'net_usem'
edges(x, sort_by = "weight", include_self = FALSE, ...)

## S3 method for class 'var_result'
edges(x, sort_by = "weight", include_self = FALSE, ...)

edges(x, ...)

## S3 method for class 'netobject'
edges(x, sort_by = "weight", include_self = FALSE, ...)

## S3 method for class 'netobject_group'
edges(x, sort_by = "weight", include_self = FALSE, ...)

## S3 method for class 'gvar_result'
edges(x, sort_by = "weight", include_self = FALSE, ...)

## S3 method for class 'net_mlvar'
edges(x, sort_by = "weight", include_self = FALSE, ...)

## S3 method for class 'net_gimme'
edges(
  x,
  sort_by = "weight",
  include_self = TRUE,
  weight = c("prop", "coef"),
  ...
)

Arguments

Value

A tidy data.frame, one row per edge.

Examples

set.seed(1)
d <- data.frame(id = 1, A = rnorm(80), B = rnorm(80), C = rnorm(80))
fit <- graphical_var(d, vars = c("A", "B", "C"), id = "id", n_lambda = 8)
edges(fit)            # tidy: network / from / to / weight

Description

Experimental. The resampling design is methodologically grounded (block bootstrap for dependent data; edge-stability summaries in the spirit of bootnet), but unlike the estimators in this package it has no external reference implementation to validate against, and its interface, defaults, and reported statistics may change in a future release.

Refits an idiographic estimator across deterministic block resamples and summarizes edge-level stability. Blocks preserve within-block time order: subject-day blocks when id and day are supplied, subjects when only id is supplied, days when only day is supplied, or consecutive row blocks for a single series. Duplicate blocks receive temporary ids/day labels before fitting so lag construction never connects two sampled copies.

Usage

estimate_stability(
  data,
  vars,
  estimator = c("var", "graphical_var", "mlvar", "usem", "gimme"),
  id = NULL,
  day = NULL,
  beep = NULL,
  n_resamples = 100L,
  resample = c("block", "split_half"),
  block_size = NULL,
  threshold = 1e-08,
  seed = NULL,
  keep_fits = FALSE,
  ...
)

Arguments

Value

A stability_result with $stability edge statistics, $original fit, $resample_edges, $failures, and $config.

See Also

build_var(), graphical_var()

Examples

set.seed(1)
d <- data.frame(id = 1, day = rep(1:4, each = 12),
                beep = rep(1:12, 4),
                A = rnorm(48), B = rnorm(48), C = rnorm(48))
st <- estimate_stability(d, vars = c("A", "B", "C"), id = "id",
                         day = "day", beep = "beep",
                         n_resamples = 5, seed = 1)
head(st$stability)

Description

Returns a one-row-per-edge data.frame with node labels, for any netobject / cograph_network (or a gvar_result constituent).

Usage

extract_edges(model, sort_by = "weight", include_self = FALSE)

Arguments

Value

A data.frame with columns from, to, weight.

Description

Estimate a graphical vector autoregressive (GVAR) model from time series or panel data. Jointly estimates a sparse temporal network (L1-penalized VAR coefficients) and a sparse contemporaneous network (graphical lasso on residuals) using EBIC model selection over a lambda grid.

Usage

graphical_var(
  data,
  vars,
  id = NULL,
  day = NULL,
  beep = NULL,
  lags = 1L,
  n_lambda = 50L,
  gamma = 0.5,
  scale = TRUE,
  center_within = TRUE,
  lambda_min_ratio = 0.05,
  lambda_min_kappa = NULL,
  lambda_min_beta = NULL,
  penalize_diagonal = TRUE,
  lambda_beta = NULL,
  lambda_kappa = NULL,
  regularize_mat_beta = NULL,
  regularize_mat_kappa = NULL,
  maxit_in = 100L,
  maxit_out = 100L,
  delete_missings = TRUE,
  likelihood = c("unpenalized", "penalized"),
  ebic_tol = 1e-04,
  mimic = "current",
  verbose = FALSE,
  min_obs = NULL,
  subject = NULL
)

Arguments

Details

This is a clean-room reimplementation of the Rothman/Epskamp two-step estimator that is numerically equivalent to graphicalVAR::graphicalVAR(): identical data preparation (global scaling, optional within-person centering, intercept column, lag-1 construction within id/day blocks), identical lambda grids (generate_lambdas), the coupled MRCE beta-update / glasso kappa-update loop, the unpenalized-likelihood EBIC, and the same tie-broken model selection. On well-conditioned data it agrees with graphicalVAR to machine precision (~1e-11); on rank-deficient data (n close to the number of parameters) it agrees to the MRCE inner-solver tolerance (~1e-3), which graphicalVAR shares.

Value

A list of class gvar_result containing:

beta

Temporal coefficient matrix, outcome x (intercept + predictors), in graphicalVAR's convention.

temporal

The p x p temporal slice beta[, -1] as [outcome, predictor] (intercept dropped).

kappa

Precision matrix (p x p, symmetric).

PCC

Partial contemporaneous correlations -cov2cor(kappa), diagonal zeroed.

PDC

Partial directed correlations.

contemporaneous

Alias for PCC.

labels

Variable names.

n_obs

Number of valid lag-pair observations.

lambda_beta, lambda_kappa

Selected penalties.

gamma, EBIC

EBIC gamma used and the selected EBIC.

References

Epskamp, S., Waldorp, L. J., Mottus, R., & Borsboom, D. (2018). The Gaussian Graphical Model in Cross-Sectional and Time-Series Data. Multivariate Behavioral Research, 53(4), 453-480.

Rothman, A. J., Levina, E., & Zhu, J. (2010). Sparse multivariate regression with covariance estimation. JCGS, 19(4), 947-962.

Description

Applies graphical_var() to each subject separately, returning one person-specific network per individual — the idiographic "all individuals" workflow. Subjects that cannot be fit (too few lag pairs after listwise deletion) are dropped with a warning.

Usage

graphical_var_each(
  data,
  vars,
  id,
  day = NULL,
  beep = NULL,
  min_obs = NULL,
  ...
)

Arguments

Value

A named list of gvar_result objects (class gvar_list), one element per subject, named by subject id.

Description

matrices() is the matrix-oriented companion to summary() and edges(). It returns the core estimated matrices invisibly and prints each matrix compactly with rounding, so users can inspect coefficients without digging through object internals.

Usage

matrices(x, ...)

## Default S3 method:
matrices(x, digits = 3, ...)

## S3 method for class 'cograph_network'
matrices(x, digits = 3, ...)

## S3 method for class 'netobject'
matrices(x, digits = 3, ...)

## S3 method for class 'netobject_group'
matrices(x, digits = 3, ...)

## S3 method for class 'gvar_result'
matrices(x, digits = 3, ...)

## S3 method for class 'var_result'
matrices(x, digits = 3, ...)

## S3 method for class 'net_mlvar'
matrices(x, digits = 3, ...)

## S3 method for class 'net_usem'
matrices(x, digits = 3, ...)

## S3 method for class 'net_gimme'
matrices(x, digits = 3, ...)

## S3 method for class 'preprocess_audit'
matrices(x, digits = 3, ...)

## S3 method for class 'rolling_var_result'
matrices(x, fit = 1L, digits = 3, ...)

## S3 method for class 'rolling_gvar_result'
matrices(x, fit = 1L, digits = 3, ...)

## S3 method for class 'stability_result'
matrices(x, digits = 3, ...)

## S3 method for class 'model_comparison'
matrices(x, fit = 1L, digits = 3, ...)

Arguments

Value

Invisibly, a named list of matrices.

Description

One row per node per network with strength (sum of absolute incident edge weights) and, for directed networks, out_strength / in_strength (NA for undirected). Self-loops are excluded.

Usage

## S3 method for class 'net_usem'
nodes(x, ...)

## S3 method for class 'var_result'
nodes(x, ...)

nodes(x, ...)

## S3 method for class 'netobject'
nodes(x, ...)

## S3 method for class 'netobject_group'
nodes(x, ...)

## S3 method for class 'gvar_result'
nodes(x, ...)

## S3 method for class 'net_mlvar'
nodes(x, ...)

## S3 method for class 'net_gimme'
nodes(x, ...)

Arguments

Value

A tidy data.frame.

Description

Draws a GIMME result the way the gimme package does: a single p-node network where dashed edges are lag-1 (temporal) and solid edges are lag-0 (contemporaneous), edge width is the proportion of subjects that have the path, black edges are group-level paths and grey edges individual-level, and autoregression shows as a dashed self-loop. Rendered with cograph::splot(), so a lag and a contemporaneous effect between the same pair are drawn as two parallel edges.

Usage

plot_gimme(
  x,
  weight = c("prop", "coef"),
  group_color = "black",
  individual_color = "grey60",
  layout = "circle",
  curvature = 0.25,
  edge_scale = 5,
  ...
)

Arguments

Value

Invisibly, the mixed cograph_network object that was plotted.

See Also

as_netobject() for the matrix view.

Examples

set.seed(1)
panel <- data.frame(
  id = rep(1:5, each = 30),
  t  = rep(seq_len(30), 5),
  A  = rnorm(150), B = rnorm(150), C = rnorm(150)
)
gm <- build_gimme(panel, vars = c("A", "B", "C"), id = "id", time = "t")
plot_gimme(gm)

Description

S3 plot() methods that render any idiographic result with cograph::splot(). Call plot(fit) to draw the full result (every network panel) or pass layer to draw a single network – "temporal", "contemporaneous", "between" (mlVAR), or "residual_cov" (uSEM) – without indexing into the object.

Usage

## S3 method for class 'var_result'
plot(x, layer = NULL, ...)

## S3 method for class 'gvar_result'
plot(x, layer = NULL, ...)

## S3 method for class 'var_bayes_result'
plot(x, layer = NULL, ...)

## S3 method for class 'net_mlvar'
plot(x, layer = NULL, ...)

## S3 method for class 'net_usem'
plot(x, layer = NULL, ...)

## S3 method for class 'net_gimme'
plot(x, layer = NULL, weight = c("prop", "coef"), ...)

## S3 method for class 'var_list'
plot(x, subject = 1L, layer = NULL, ...)

## S3 method for class 'gvar_list'
plot(x, subject = 1L, layer = NULL, ...)

## S3 method for class 'rolling_var_result'
plot(x, fit = 1L, layer = NULL, ...)

## S3 method for class 'rolling_gvar_result'
plot(x, fit = 1L, layer = NULL, ...)

## S3 method for class 'stability_result'
plot(x, layer = NULL, ...)

Arguments

Value

Invisibly, the object that was plotted (a cograph/ggplot object).

Examples

set.seed(1)
d <- data.frame(id = 1, A = rnorm(80), B = rnorm(80), C = rnorm(80))
fit <- build_var(d, vars = c("A", "B", "C"), id = "id")
plot(fit)
plot(fit, layer = "temporal")

Description

Prints a concise summary of estimator, successful rolling splits, prediction count, and overall RMSE.

Usage

## S3 method for class 'forecast_result'
print(x, ...)

Arguments

Value

x, invisibly.

Description

Print a list of per-subject graphical VARs

Usage

## S3 method for class 'gvar_list'
print(x, ...)

Arguments

Value

x, invisibly.

Description

Print Method for gvar_result

Usage

## S3 method for class 'gvar_result'
print(x, digits = 2, ...)

Arguments

Value

The input object, invisibly.

Description

Prints a concise summary of requested estimators, successful fits, and failures in a model comparison report.

Usage

## S3 method for class 'model_comparison'
print(x, ...)

Arguments

Value

x, invisibly.

Description

Print Method for net_gimme

Usage

## S3 method for class 'net_gimme'
print(x, digits = 2, ...)

Arguments

Value

The input object, invisibly.

Examples

set.seed(1)
panel <- data.frame(
  id = rep(1:5, each = 20),
  t  = rep(seq_len(20), 5),
  A  = rnorm(100), B = rnorm(100), C = rnorm(100)
)
gm <- build_gimme(panel, vars = c("A","B","C"), id = "id", time = "t")
print(gm)

Description

Print method for net_mlvar

Usage

## S3 method for class 'net_mlvar'
print(x, digits = 2, ...)

Arguments

Value

Invisibly returns x.

Examples

set.seed(1)
n_id <- 8; n_t <- 30; vars <- c("A", "B", "C")
rows <- lapply(seq_len(n_id), function(i) {
  m <- as.data.frame(matrix(rnorm(n_t * 3), ncol = 3))
  names(m) <- vars
  m$id <- i; m$day <- 1L; m$beep <- seq_len(n_t)
  m
})
d <- do.call(rbind, rows)
fit <- build_mlvar(d, vars = vars, id = "id", day = "day", beep = "beep")
print(fit)
summary(fit)

Description

Print method for net_mlvar_bayes

Usage

## S3 method for class 'net_mlvar_bayes'
print(x, digits = 2, ...)

Arguments

Value

Invisibly returns x.

Description

Print method for uSEM fits

Usage

## S3 method for class 'net_usem'
print(x, digits = 2, ...)

Arguments

Value

x, invisibly.

Description

Prints a concise summary of variables, ordered rows, retained lag pairs, trend flags, high-AR flags, and zero-variance flags.

Usage

## S3 method for class 'preprocess_audit'
print(x, ...)

Arguments

Value

x, invisibly.

Description

Prints a concise summary of subject count, window count, and variable count for a rolling graphical VAR result.

Usage

## S3 method for class 'rolling_gvar_result'
print(x, ...)

Arguments

Value

x, invisibly.

Description

Prints a concise summary of subject count, window count, and variable count for a rolling VAR result.

Usage

## S3 method for class 'rolling_var_result'
print(x, ...)

Arguments

Value

x, invisibly.

Description

Prints a concise summary of estimator, resampling type, successful refits, and the size of the edge-stability table.

Usage

## S3 method for class 'stability_result'
print(x, ...)

Arguments

Value

x, invisibly.

Description

Print method for var_bayes_result

Usage

## S3 method for class 'var_bayes_result'
print(x, digits = 2, ...)

Arguments

Value

Invisibly x.

Description

Prints a concise summary of subject count, variable count, lag-pair counts, and temporal edge counts for a build_var_each() result.

Usage

## S3 method for class 'var_list'
print(x, ...)

Arguments

Value

x, invisibly.

Description

Print method for ordinary VAR fits

Usage

## S3 method for class 'var_result'
print(x, digits = 2, ...)

Arguments

Value

x, invisibly.

Description

Fits graphical_var() over ordered, overlapping windows within each subject. This is the time-varying graphical VAR companion to rolling_var(): every window uses graphical VAR's lag construction, EBIC/penalty settings, and tidy coefficient access, then returns one coefficient table per window.

Usage

rolling_graphical_var(
  data,
  vars,
  id = NULL,
  day = NULL,
  beep = NULL,
  window_size,
  step = 1L,
  scale = TRUE,
  center_within = TRUE,
  delete_missings = TRUE,
  min_obs = NULL,
  subject = NULL,
  keep_fits = FALSE,
  ...
)

Arguments

Value

A rolling_gvar_result with $estimates, $windows, $failures, and optionally $fits. $estimates is a tidy coefficient table with subject/window metadata plus network, from, to, and weight.

See Also

graphical_var(), rolling_var()

Examples

set.seed(1)
d <- data.frame(id = 1, day = rep(1:5, each = 20),
                beep = rep(1:20, 5),
                A = rnorm(100), B = rnorm(100), C = rnorm(100))
tv <- rolling_graphical_var(d, vars = c("A", "B", "C"), id = "id",
                            day = "day", beep = "beep",
                            window_size = 50, step = 25,
                            scale = FALSE, n_lambda = 5)
head(tv$estimates)

Description

Fits build_var() over ordered, overlapping windows within each subject. This is a simple time-varying idiographic baseline: every window uses the same lag construction, scaling, within-person centering, and tidy coefficient access as build_var(), but returns one coefficient table per window.

Usage

rolling_var(
  data,
  vars,
  id = NULL,
  day = NULL,
  beep = NULL,
  window_size,
  step = 1L,
  scale = TRUE,
  center_within = TRUE,
  delete_missings = TRUE,
  min_obs = NULL,
  subject = NULL,
  keep_fits = FALSE
)

Arguments

Value

A rolling_var_result with $estimates, $windows, $failures, and optionally $fits. $estimates is a tidy coefficient table with subject/window metadata plus network, from, to, and weight.

See Also

build_var(), audit_preprocess()

Examples

set.seed(1)
d <- data.frame(id = 1, day = rep(1:5, each = 20),
                beep = rep(1:20, 5),
                A = rnorm(100), B = rnorm(100), C = rnorm(100))
tv <- rolling_var(d, vars = c("A", "B", "C"), id = "id",
                  day = "day", beep = "beep",
                  window_size = 40, step = 20, scale = FALSE)
head(tv$estimates)

Description

The self-regulated learning (SRL) experience-sampling data used in the Learning Analytics Methods book, Chapter 20 (Vector Autoregression). Each of 36 students reported nine self-regulated-learning indicators once per study occasion for 156 occasions, giving a balanced person-by-time panel suitable for the idiographic time-series methods in this package.

Usage

srl

Format

A data.frame with 5616 rows and 11 columns:

name

Student name (36 unique students).

day

Within-person occasion index, 1-156.

efficacy

Self-efficacy.

value

Task value.

planning

Planning.

monitoring

Monitoring.

effort

Effort regulation.

control

Control of learning.

help

Help seeking.

social

Social support.

organizing

Organizing.

Details

The columns have already been tidied for modelling: rows are ordered by name then day, and day is a within-person occasion index (1-156) you can pass as the time argument to build_usem() and build_gimme(). No further ordering, indexing, or column selection is needed before fitting a model.

Source

Learning Analytics Methods, Book 2, Chapter 20 (VAR): https://lamethods.org/book2/chapters/ch20-var/ch20-var.html. Original data: https://github.com/lamethods/data2/raw/main/srl/srl.RDS.

Examples

data(srl)
summary(srl)
head(srl)

Description

Summary Method for gvar_result

Usage

## S3 method for class 'gvar_result'
summary(object, ...)

Arguments

Value

A tidy data.frame of per-network metrics: one row per network (temporal, contemporaneous) with n_nodes, n_edges, density, mean_abs_weight, n_positive, n_negative. Use edges(object) / coefs(object) for the estimates and nodes(object) for node strengths.

Description

Summary Method for net_gimme

Usage

## S3 method for class 'net_gimme'
summary(object, ...)

Arguments

Value

A tidy data.frame of per-network metrics (one row per network: temporal, contemporaneous), with n_edges/density/etc. computed from the proportion-of-subjects networks. Per-subject fit indices are in object$fit; coefs(object) gives the per-person estimates, edges(object) the tidy edge list, and nodes(object) node strengths.

Examples

set.seed(1)
panel <- data.frame(
  id = rep(1:5, each = 20),
  t  = rep(seq_len(20), 5),
  A  = rnorm(100), B = rnorm(100), C = rnorm(100)
)
gm <- build_gimme(panel, vars = c("A","B","C"), id = "id", time = "t")
summary(gm)

Description

Summary method for net_mlvar

Usage

## S3 method for class 'net_mlvar'
summary(object, ...)

Arguments

Value

A tidy data.frame of per-network metrics (one row per network: temporal, contemporaneous, between). Use coefs(object) for the fixed-effect coefficient table, edges(object) for the edge list, and nodes(object) for node strengths.

Examples

set.seed(1)
n_id <- 8; n_t <- 30; vars <- c("A", "B", "C")
rows <- lapply(seq_len(n_id), function(i) {
  m <- as.data.frame(matrix(rnorm(n_t * 3), ncol = 3))
  names(m) <- vars
  m$id <- i; m$day <- 1L; m$beep <- seq_len(n_t)
  m
})
d <- do.call(rbind, rows)
fit <- build_mlvar(d, vars = vars, id = "id", day = "day", beep = "beep")
print(fit)
summary(fit)

Description

Summary method for uSEM fits

Usage

## S3 method for class 'net_usem'
summary(object, ...)

Arguments

Value

A tidy per-network metrics data.frame.

Description

Summary method for var_bayes_result

Usage

## S3 method for class 'var_bayes_result'
summary(object, ...)

Arguments

Value

A tidy per-network metrics data.frame.

Description

Summary method for ordinary VAR fits

Usage

## S3 method for class 'var_result'
summary(object, ...)

Arguments

Value

A tidy per-network metrics data.frame.

Description

Experimental. The rolling-origin design follows standard time-series cross-validation practice, but unlike the estimators in this package it has no external reference implementation to validate against, and its interface, defaults, and reported metrics may change in a future release.

Performs rolling-origin one-step prediction from build_var() or graphical_var(). Each split fits the estimator on earlier blocks and predicts current variables in the next block from their lag-1 values. Scaling and within-person centering parameters are learned from the training split only, then applied to the assessment split before prediction.

Usage

validate_forecast(
  data,
  vars,
  estimator = c("var", "graphical_var"),
  id = NULL,
  day = NULL,
  beep = NULL,
  initial = NULL,
  assess = 1L,
  step = 1L,
  n_splits = NULL,
  block_size = NULL,
  scale = TRUE,
  center_within = TRUE,
  delete_missings = TRUE,
  keep_fits = FALSE,
  ...
)

Arguments

Value

A forecast_result with $predictions, $metrics, $splits, $failures, and optionally $fits.

See Also

build_var(), graphical_var(), estimate_stability()

Examples

set.seed(1)
d <- data.frame(id = 1, day = rep(1:5, each = 12),
                beep = rep(1:12, 5),
                A = rnorm(60), B = rnorm(60), C = rnorm(60))
fc <- validate_forecast(d, vars = c("A", "B", "C"), id = "id",
                        day = "day", beep = "beep",
                        initial = 3, n_splits = 2, scale = FALSE)
fc$metrics