Package index • lame

Model Fitting (MCMC)

Core Bayesian MCMC estimators for AME and LAME models

ame(): AME model fitting routine
lame(): AME model fitting routine for longitudinal relational data
lame-package: Longitudinal Additive and Multiplicative Effects Models for Networks
ame_parallel(): Run AME model with multiple parallel chains
lame_parallel(): Run LAME (longitudinal AME) with multiple parallel chains
lame_multi(): Multi-panel lame() with shared coefficients
lame_resume(): Resume a lame() MCMC run from a checkpoint
ame_options(): AME model fitting options

Forecasting and Temporal Diagnostics

Forecasts, counterfactuals, and diagnostics for dynamic fits

gof_temporal(): Posterior-predictive temporal-trend test
detect_change_point(): Detect potential change points in a dynamic_beta posterior path
lfo(): Exact rolling-origin leave-future-out cross-validation
forecast_pit(): Probability-integral-transform calibration check for h-step forecasts
rhat_dynamic_beta(): Multivariate split-R-hat for dynamic_beta coefficient paths
dynamic_beta_prior_summary(): Summarise the implied prior on a time-varying coefficient path
prediction_draws_long(): Long-format draws of the linear predictor for marginaleffects-style use
per_actor_slopes(): Post-MCMC per-actor time-varying slopes
als_dynamic_beta(): Penalised ALS time-varying coefficient estimate

Model Comparison and Posterior Diagnostics

Information criteria, posterior draws, and held-out evaluation

loo() waic(): Generic dispatcher for loo / waic on ame / lame fits
loo(<ame>) loo(<lame>) loo(<ame_als>): Approximate leave-one-out cross-validation for AME / LAME fits
waic(<ame>) waic(<lame>) waic(<ame_als>): WAIC for AME / LAME fits
as_draws(): Generic dispatcher for posterior::as_draws on lame fits
as_draws(<ame>) as_draws(<lame>) as_draws(<ame_als>): Convert an AME / LAME fit to a posterior draws object
prior_summary(): Print the priors used by an AME / LAME / ame_als fit
evaluate_heldout(): Held-out predictive evaluation for an ame / lame fit
read_log_lik(): Read the per-iteration log-lik matrix back from on-disk chunks
snap_index_draws(): Extract posterior draws of snap indices
snap_index_summary(): Summarize posterior snap indices
snap_category_summary(): Summarize snap indices by actor category
snap_rank_summary(): Summarize posterior rank uncertainty for snap years

Tidyverse Integration

broom and ggplot2 methods for fitted models

tidy(): S3 generic for tidy
tidy(<ame>) tidy(<lame>): Tidy method for fitted ame / lame objects
tidy(<ame_als>) tidy(<lame_als>): Tidy method for fitted ame_als / lame_als objects
tidy(<boot_ame>): Tidy method for a standalone bootstrap object (boot_ame)
glance(): S3 generic for glance
glance(<ame>) glance(<lame>): Glance method for fitted ame / lame objects
glance(<ame_als>) glance(<lame_als>): Glance method for fitted ame_als / lame_als objects
autoplot(<lame>) autoplot(<ame>): Ribbon plot of time-varying coefficients (or coefplot for static fits)
autoplot(<ame_als>) autoplot(<lame_als>): autoplot method for ALS fits

Fast Estimator (MCMC-free)

Iterative block coordinate descent point estimator with bootstrap uncertainty

ame_als(): Fast (MCMC-free) AME estimation for a cross-sectional network
lame_als(): Fast (MCMC-free) AME estimation for a longitudinal network
lame_snap_als(): Fast approximate dynamic snap-shift AME estimator
als_start_vals(): Convert an ALS fit to MCMC starting values
ame_als_bootstrap() boot_ame(): Bootstrap uncertainty for the fast AME estimator
ame_als_refit(): Refit a fast AME model with a warm start
sampler_describe(): Describe the estimator behind a fitted object

S3 Methods

Standard R methods for model objects

coef(<ame>) coef(<lame>): Extract model coefficients from AME model
vcov(<ame>) vcov(<lame>): Posterior covariance of AME model coefficients
confint(<ame>) confint(<lame>): Bayesian credible intervals for AME model parameters
predict(<ame>): Predict method for AME models
predict(<lame>): Predict method for LAME models
fitted(<ame>): Extract fitted values from AME model
fitted(<lame>): Extract fitted values from LAME model
residuals(<ame>): Extract residuals from AME model
residuals(<lame>): Extract residuals from LAME model
simulate(<ame>): Simulate networks from a fitted AME model
simulate(<lame>): Simulate longitudinal networks from a fitted LAME model
summary(<ame>): Summary of an AME object
summary(<lame>): Summary of a LAME object
print(<ame>): Print method for AME model objects
print(<lame>): Print method for LAME objects
print(<summary.ame>): Print method for summary.ame objects
print(<summary.lame>): Print method for summary.lame objects
plot(<ame>): Simple diagnostic plot for AME model fit
plot(<lame>): Plot diagnostics for a LAME model fit
coef(<ame_als>): Extract coefficients from a fast AME fit
vcov(<ame_als>): Sandwich covariance for the regression coefficients of a fast AME fit
confint(<ame_als>): Confidence intervals for a fast AME fit
fitted(<ame_als>): Extract fitted values from a fast AME fit
residuals(<ame_als>): Residuals from a fast AME fit
predict(<ame_als>): Predictions from a fast AME fit
logLik(<ame_als>): Log-likelihood is not defined for a fast AME fit
plot(<ame_als>): Plot the convergence of a fast AME fit
print(<ame_als>): Print an ame_als object
summary(<ame_als>): Summarize an ame_als object
print(<boot_ame>): Print bootstrap results for a fast AME fit
summary(<boot_ame>) print(<summary.boot_ame>): Summarize bootstrap results for a fast AME fit
confint(<boot_ame>): Confidence intervals from a fast AME bootstrap
vcov(<boot_ame>): Bootstrap covariance of the regression coefficients
coef(<boot_ame>): Point estimates from a fast AME bootstrap
fitted(<boot_ame>) residuals(<boot_ame>): fitted/residuals are not defined for a bootstrap object
formula(<ame>) formula(<lame>): formula() is not defined for an ame() / lame() fit
logLik(<ame>) logLik(<lame>): Log-likelihood is not directly exposed for ame() / lame() fits
nobs(<ame>) nobs(<lame>): Number of observed dyads in an AME / LAME fit
nobs(<ame_als>): Number of observed dyads in an ame_als fit
update(<ame>) update(<lame>): Update an AME / LAME fit
update(<ame_als>) update(<lame_als>): Update an ame_als / lame_als fit
simulate(<ame_als>): Simulate networks from a fitted ame_als model
ab_plot.ame_als(): Additive-effects plot for an ame_als fit
gof_plot.ame_als(): Goodness-of-fit check for an ame_als fit
coef(<als_dynamic_beta>): Extract beta path from a penalised-ALS object
print(<als_dynamic_beta>): Print method for penalised ALS time-varying beta
print(<gof_temporal>): Print method for gof_temporal output
print(<lame_multi>): Print method for lame_multi
print(<lfo_lame>): Print method for lfo() results
print(<per_actor_slopes>): Print method for per_actor_slopes
print(<summary.ame_als>): Print a fast AME summary

Visualization

Plotting functions for diagnostics and model exploration

trace_plot(): MCMC trace plots and density plots for AME/LAME model parameters
gof_plot(): Visualize goodness-of-fit statistics for AME and LAME models
ab_plot(): Visualize sender and receiver random effects
uv_plot(): Visualize multiplicative effects (latent factors) from AME models

Goodness of Fit

Posterior predictive checks and GOF statistics

gof(): Compute GOF statistics from saved posterior samples
gof_stats(): Goodness of fit statistics
gof_stats_unipartite(): Goodness of fit statistics for unipartite networks
gof_stats_bipartite(): Goodness of fit statistics for bipartite networks
compute_gof_bipartite(): Compute GOF statistics for bipartite networks

Latent Space

Extract and align latent positions

latent_positions(): Extract latent positions as a tidy data frame
procrustes_align(): Procrustes alignment of latent positions across time

Posterior Utilities

Tools for working with posterior samples

simulate_posterior(): Simulate posterior distributions from fitted AME model
posterior_quantiles(): Extract posterior quantiles for model components
posterior_options(): Options for saving posterior samples during MCMC
reconstruct_EZ() reconstruct_UVPM(): Reconstruct EZ and UVPM matrices from AME model output
combine_ame_chains(): Combine multiple AME chains
compute_mcmc_diagnostics(): Compute MCMC convergence diagnostics for multiple chains

Simulation

Simulate network data from fitted models

print(<ame.sim>) print(<lame.sim>): Print methods for AME and LAME simulation objects
summary(<ame.sim>): Summary method for AME simulations
summary(<lame.sim>): Summary method for LAME simulations
simY_nrm(): Simulate a normal relational matrix
simY_bin(): Simulate a binary relational matrix from latent means
simY_ord(): Simulate an ordinal relational matrix
simY_pois(): Simulate a Poisson relational matrix
simY_frn(): Simulate a relational matrix under a fixed rank nomination scheme
simZ(): Simulate a latent relational Gaussian array

Data Manipulation

Convert between network data formats and build covariates

el2sm(): Edgelist to sociomatrix
sm2el(): Sociomatrix to edgelist
array_to_list(): Convert array to list.
list_to_array(): Convert list to array
list_to_array_bipartite(): Convert bipartite list data to array format
as_lame_y(): Convert a graph object to a lame-ready adjacency matrix
nodematch() absdiff() nodefactor(): ERGM-style covariate helpers for ame() / lame()
check_format(): Validate input data format for lame function
zscores(): rank-based z-scores

Memory and Performance

Manage memory usage in large models

compact_ame(): Optimize AME model output for memory efficiency
ame_memory_usage(): Calculate memory usage of AME model components
ame_memory_settings(): Display memory usage information for AME models

MCMC Samplers

Gibbs sampling functions for latent variables, regression coefficients, and variance components

rZ_nrm_fc(): Simulate missing values in a normal AME model
rZ_bin_fc(): Draw latent Z for the binary probit model
rZ_ord_fc(): Full-conditional latent draw for the ordinal family
rZ_pois_fc(): Gibbs update for latent variable in a Poisson AME model
rZ_cbin_fc(): Draw latent Z for censored-binary nomination data
rZ_frn_fc(): Draw the latent Z matrix for fixed-rank-nomination data
rbeta_ab_fc(): Joint Gibbs update of regression and additive effects (single relation)
rbeta_ab_rep_fc(): Gibbs update of regression coefficients and additive effects for replicated relational data
rSab_fc(): Gibbs update for additive effects covariance
rSuv_fc(): Gibbs update for multiplicative effects covariance
rUV_fc(): Gibbs sampling of U and V
rUV_rep_fc(): Gibbs sampling of U and V from replicated relational data
rUV_sym_fc(): Gibbs sampling of U and V
rUV_dynamic_fc(): Gibbs sampling of dynamic U and V with AR(1) evolution
rUV_dynamic_fc_cpp(): Update dynamic latent positions using AR(1) process
rUV_dynamic_snap_fc(): Gibbs sampling of dynamic U and V with snap-shift dynamics
rUV_dynamic_snap_fc_cpp(): Update dynamic latent positions with snap-shift model selection
rUV_dynamic_t_fc(): Gibbs sampling of dynamic U and V with heavy-tailed (Student-t) innovations
rUV_dynamic_t_fc_cpp(): Update dynamic latent positions with heavy-tailed (Student-t) AR(1) innovations
raSab_bin_fc(): Sample additive row effects and their covariance for the binary family
raSab_cbin_fc(): Simulate a and Sab from full conditional distributions under the cbin likelihood
raSab_frn_fc(): Simulate a and Sab from full conditional distributions under frn likelihood
rrho_mh(): Metropolis-Hastings update for the within-dyad correlation
rs2_fc(): Gibbs update for dyadic variance
rs2_rep_fc(): Full-conditional draw of the dyadic variance for replicated relational data
sample_ab_bipartite(): Sample additive effects for bipartite networks
sample_dynamic_ab_cpp(): Sample dynamic additive effects with AR(1) evolution
sample_rho_ab_cpp(): Sample AR(1) parameter for dynamic additive effects
sample_rho_uv(): Sample AR(1) parameter for dynamic latent factors
sample_sigma_ab_cpp(): Sample innovation variance for dynamic additive effects
sample_sigma_uv(): Sample innovation variance for dynamic latent factors
init_dynamic_ab_cpp(): Initialize dynamic additive effects with AR(1) structure
init_dynamic_positions(): Initialize dynamic latent positions with AR(1) structure
update_variances_bipartite(): Update variance parameters for bipartite
rUV_dynamic_bip_fc_cpp(): Bipartite dynamic UV Gibbs update
rZ_nrm_batch_cpp(): Batch normal Z sampling across all time periods
rZ_bin_bip_batch_cpp(): Batch binary Z sampling across all time periods (bipartite, rho=0)
compute_XtX_Xty_bip_cpp(): Compute X'X and X'y for bipartite covariate regression
Xbeta_bip_cpp(): Compute Xbeta product for bipartite networks
rbeta_ab_bip_gibbs_cpp(): Full bipartite Gibbs update for beta, a, b
sample_beta_dynamic_cpp(): Sample the dynamic-block beta path via FFBS
sample_beta_static_cpp(): Sample the static-block beta conditional on the dynamic path
sample_rho_beta_cpp(): Sample the AR(1) rho for each dynamic block
sample_sigma_beta_cpp(): Sample the AR(1) innovation sigma for each dynamic block
get_EZ_dynamic_beta_cpp(): Compute EZ when beta is time-varying

Internal Helpers

Design matrix construction and low-level utilities

init_bipartite_startvals(): Bipartite network helper functions
design_array(): Assemble the dyadic design socioarray for an AME model
design_array_listwisedel(): Computes the design socioarray of covariate values
get_design_rep(): Create design array for replicate data
get_fit_object(): Get fitted object from MCMC results
get_start_vals(): Get fitted object from MCMC results
precomputeX(): Precompute design-array cross-product summaries
Xbeta(): Linear combinations of submatrices of an array
mhalf(): Symmetric square root of a matrix
rmvnorm(): Simulation from a multivariate normal distribution
rwish(): Simulation from a Wishart distribution

Datasets

Example network datasets

YX_nrm: normal relational data and covariates
YX_bin: binary relational data and covariates
YX_bin_list: Synthetic longitudinal binary relational data, list-form (latent-scale)
YX_bin_long: synthetic longitudinal binary relational data (latent-scale)
YX_ord: ordinal relational data and covariates
YX_cbin: Censored binary nomination data and covariates
YX_frn: Fixed rank nomination data and covariates
IR90s: International relations in the 90s
coldwar: Cold War data
comtrade: Comtrade data
lazegalaw: Lazega's law firm data
dutchcollege: Dutch college data
sampsonmonks: Sampson's monastery data
sheep: Sheep dominance data
addhealthc3: AddHealth community 3 data
addhealthc9: AddHealth community 9 data
vignette_data: TIES sanctions data for vignettes
Y: Relational matrix
Xdyad: Dyadic covariates
Xrow: Row covariates
Xcol: Column covariates