Guides¶
Task-oriented pages: everything that is not "which model should I use".
-
Configuration and ground truth
simulate()returns the coefficients, latent times and cure status the frame cannot show — what a simulation study is actually about. -
The five hazard families every sampler draws from, and how to supply one of your own.
-
The two built-in mechanisms, hitting a target event rate, and applying a censoring distribution of your own.
-
Three different covariate schemes across the model families, and how to control each.
-
Event counts, follow-up, quality checks, and comparing several datasets at once.
-
Kaplan-Meier curves, stratified curves, hazard comparisons, covariate effects.
-
CSV, JSON, Feather and RDS — and what a round trip loses.
-
lifelines, scikit-survival, scikit-learn and R.
-
Generating and plotting without writing Python.
Runnable notebooks¶
Three worked notebooks live in the repository under examples/notebooks/, and
open in a browser with no install through Binder:
| Notebook | Opens in Binder |
|---|---|
Illness-death, intervals (cmm) |
|
Illness-death, panel (thmm) |
|
Time-dependent covariates (tdcm) |
The same directory also holds plain scripts — run_cphm.py, run_aft.py,
run_competing_risks.py and others — that can be run directly from a checkout.