Package Overview | Migrating from amen | Application: Dutch College Friendships | Fitting the Model | Interpreting the Results | Checking Convergence | Bayesian-ecosystem diagnostics: posterior::as_draws() | Goodness of Fit | Posterior-predictive temporal-trend test: gof_temporal() | Latent Space | Extracting Latent Positions | Tidy-data round-trip via netify | Entry: tidy edgelist → lame() | Exit: tidy(), autoplot(), prediction_draws_long() | Coefficient tables: glance() + tidy() -> modelsummary / gt / kableExtra | Model comparison: loo::loo_compare() as a one-liner | Reproducibility: sessionInfo() and renv.lock | What's Next?

Why Network Models? | The Data | Building Covariates | Fitting the Model | A note for ERGM users: what AME assumes | What Did We Find? | Did the Sampler Converge? | Numerical convergence diagnostics: posterior::as_draws() | Between-chain $\hat R$: ame_parallel(n_chains = 4) | Other quick checks | Model comparison: loo() and loo_compare() | Does the Model Fit the Data? | Visualizing the Latent Space | Who Are the Most Social / Popular Students? | Making Predictions | Held-out link prediction | Simulating from the Model | Practical Tips | What's Next? | References

Introduction | The Bipartite AME Model | Supported families in bipartite mode | Cross-Sectional Analysis with ame() | Coming from long-format data? | Simulating Data | Fitting the Model | Visualizing the Latent Space | Longitudinal Analysis with lame() | Simulating Longitudinal Data | Static Model | Dynamic Model | Time-varying Interaction Matrix (dynamic_G) | Visualizing Temporal Evolution | Convergence Diagnostics | Comparing Static and Dynamic Models | Choosing Latent Dimensions | Practical Guidance | Extracting Positions for Custom Analysis

Introduction | What Are Dynamic Effects? | The Static Baseline | Dynamic Latent Positions (dynamic_uv = TRUE) | Dynamic Additive Effects (dynamic_ab = TRUE) | Dynamic Regression Coefficients (dynamic_beta) | A worked dynamic_beta example | Choosing a prior family: dynamic_beta_kind | Hierarchical pooling: dynamic_beta_pool | Per-actor time-varying slopes: dynamic_beta_per_actor | Diagnosing your dynamic fit | Temporal-trend posterior-predictive check: gof_temporal() | Inspecting the priors actually in effect: prior_summary() | Multi-chain $\hat R$ via lame_parallel() and rhat_dynamic_beta() | Forecasting | Visualising the coefficient path: autoplot.lame | LOO / WAIC via save_log_lik = TRUE | Prior Specifications | Genuine Dynamics: What They Look Like | What Happens Without Temporal Structure? | Visualizing the Null Case | Comparing Model Specifications | When to Use Dynamic Effects | Dealing with Rotational Indeterminacy | References

When to use the fast estimator | Wall-clock benchmark | What ALS can do vs. what only MCMC does | A first fit | Choosing the rank R | Non-normal families | Uncertainty | ALS bootstrap vs MCMC posterior on the same data | Longitudinal data | Summary

What this vignette covers | Step 1: Fit | Step 2: Forecast h periods ahead | Visualising the forecast | Poisson exposure offsets via period_exposure | Step 3: Counterfactual forecasts | Step 4: Visualise the coefficient path | Step 5: Forecast diagnostics | Watching it fail: a sparse, short, drifting panel | Probability-integral-transform (PIT) calibration | Comparing two fits with loo_compare() | Memory-conscious long runs | See also

What Is lame? | Coming from a tidy edgelist? Use netify | A 5-Minute Example | What Can You Do with a Fitted Model? | Checking Your Model | Visualizing Network Structure | Cross-Sectional Models | Dynamic Effects | Bipartite Networks | Supported Data Types | Power-User Features | Where to Go Next | References

new to networks | overview books | academic articles | networks & conflict | model-specific

the netify object: a base r object with attributes | extracting parts | tidy interop | predicates and descriptors | object-level validation | open-cohort panels with actor_pds | the actor_pds roster, step by step | density and per-period actor sets | na versus zero in weighted networks | writing a custom graph-level statistic | writing a custom actor-level statistic | reading the dyad_data nested list | writing a custom exporter (to_* function) | performance characteristics | see also

1. cross-sectional pipeline | before you fit: three sanity checks | dyadic edge covariates: the _e suffix | 2. longitudinal pipeline (per-time fits) | 3. multilayer pipeline (per-layer fits) | 4. round-tripping simulated networks back to netify | tl;dr | references

1. build | 2. explore | 3. test (basic inferential) | 4. bridge | tl;dr | a note for non-time use cases