Regression Modelling
Models in which covariates change the survival of an individual unit. The families differ in how a covariate acts: multiplying the hazard, adding to it, scaling time, or shifting the odds.
The univariate regression models are importable directly from
surpyval. For the theory see Regression Analysis; for a
narrative introduction with worked examples, see
Regression Modelling with SurPyval. Tree-based models are under
Machine Learning (beta), and regression for recurrent events under
Recurrent Event Models.
Semi-Parametric Models
No assumed baseline distribution: the shape of the baseline is left to the data, and only the covariate effect is parameterised.
Parametric Models
A fitted baseline distribution combined with a covariate function. The page below covers the proportional-hazards, accelerated-failure-time, proportional-odds, parametric additive-hazards and accelerated-life families, the fitted model they all return, and the time-varying covariate paths (step schedules and continuously varying paths) the PH, AFT, AH and PO families can be evaluated along.