Quick start

A 10-line tour of amorem: simulate a relational event stream with a known driver, then recover the driver’s coefficient from the emitted case-control table.

Numbers shown below are actual output, re-run on every release.

Install

From GitHub:

# install.packages("remotes")
remotes::install_github("franciscorichter/amorem")

Hard dependencies: Rcpp, survival (for coxph / clogit), mgcv (for smooth-effect curves). Suggested: coxme (two-axis random effects).

Simulate, then recover

A 20-actor network, 1200 events, with a single endogenous mechanism — reciprocity at strength β = 0.6:

library(amorem)
suppressPackageStartupMessages(library(survival))

set.seed(1)
cc <- simulate_relational_events(
  n_events           = 1200,
  senders            = paste0("a", 1:20),
  receivers          = paste0("a", 1:20),
  baseline_rate      = 1,
  n_controls         = 1,                 # one matched non-event per event
  endogenous_stats   = "reciprocity_count",
  endogenous_effects = c(reciprocity_count = 0.6))

fit <- clogit(event ~ reciprocity_count + strata(stratum), data = cc)
coef(fit)
#> reciprocity_count 
#>             0.542
confint(fit)
#>                       2.5 %    97.5 %
#> reciprocity_count    0.349     0.734

The 95% interval (0.35, 0.73) covers the true coefficient 0.6. The simulator emits a case-control table directly (with one sampled non-event per event when n_controls = 1), so the recovery step is a single clogit call.

For preprocessed case-control data (e.g. from eventnet) the same fit, the case-1-control degenerate-logistic variant, and smooth (TV / NL / TVNL) effects are all available through one interface, rem() — see Estimation.

Where to go next

Page	What it covers
Simulation	The five dyadic simulator modes and the hyperedge simulator for multi-actor meetings
Endogenous catalogue	The 68-stat dyadic table plus hyperedge-native subset repetition
Estimation	Case-control sampling, linear/smooth/global-covariate model comparison, GOF tests
Hyperedge models	RHEMs with set-valued sender/receiver hyperedges
Real-data analysis	Classroom, Manufacturing, frailty, smooth effects
Validation experiments	Parity tests + MLE recovery + scaling
Datasets	The three bundled REM datasets and their provenance