API reference

Generated from the source, so it is always in step with the version you have installed. For task-oriented material, start from Guides or the model pages.

Everything in gen_surv

from gen_surv import ... gives you:

Name	Kind	Page
generate	dispatcher over all twelve models	generate
simulate, SimulationConfig, SimulationResult	data with its configuration and ground truth	Simulation results
gen_cphm	Cox proportional hazards	Generators
gen_aft_log_normal, gen_aft_weibull, gen_aft_log_logistic	AFT models	Generators
gen_piecewise_exponential	piecewise constant hazard	Generators
gen_competing_risks, gen_competing_risks_weibull	competing risks	Generators
gen_mixture_cure, cure_fraction_estimate	cure models	Generators
gen_cmm	illness-death, intervals	Generators
gen_thmm	illness-death, panel	Generators
gen_tdcm	time-dependent covariates	Generators
gen_recurrent_events	recurrent events (Andersen-Gill, PWP)	Generators
gen_multistate, Transition	the engine behind cmm and thmm; any transition graph	Generators
BaselineHazard, ExponentialBaseline, WeibullBaseline, GompertzBaseline, LogLogisticBaseline, PiecewiseConstantBaseline	baseline hazard families	Baseline hazards
runifcens, rexpocens, rweibcens, rlognormcens, rgammacens	censoring samplers	Censoring
WeibullCensoring, LogNormalCensoring, GammaCensoring, CensoringModel	class-based censoring	Censoring
sample_bivariate_distribution	correlated draws	Censoring
describe_survival, plot_survival_curve, plot_hazard_comparison, plot_covariate_effect	summaries and plots	Analysis
export_dataset	write to disk	Interoperability
to_sksurv, from_sksurv	scikit-survival conversion — needs the optional dependency	Interoperability
GenSurvDataGenerator	scikit-learn estimator	Interoperability
__version__	the installed version	—

Not exported at the top level

These live one level down and are imported from their module:

Import	Purpose
from gen_surv.summary import summarize_survival_dataset, check_survival_data_quality, compare_survival_datasets	dataset summaries and quality checks — Analysis
from gen_surv.validation import ValidationError, ...	the exception hierarchy — Validation
from gen_surv.cli import app	the Typer application — Command line

Conventions across the package

Every generator returns a DataFrame. Never a numpy array, never a tuple. Shapes differ by model — see Output schemas.

seed is always last and always optional. It accepts an int, a numpy.random.Generator, or None. See Reproducibility.

Validation happens before any sampling. An invalid argument raises a subclass of ValidationError — itself a ValueError — with the offending argument, the value received, and what was expected.

Optional dependencies degrade quietly. to_sksurv and from_sksurv are simply absent when scikit-survival is not installed; importing gen_surv still works.