API

The reference for every public class and function in SurPyval, one page per area. The theory behind each area is in the Survival Analysis pages and worked examples are in the SurPyval Modelling pages; each reference page below links to both.

Reading these pages. SurPyval follows one pattern throughout: a fitter takes data through fit() (or fit_from_df() for a DataFrame) and returns a fitted model, which you then ask for survival probabilities, hazards, quantiles, bounds and plots. Most fitters are exported as ready-made instances rather than classes, so the reference documents their class, whose name usually ends in an underscore: surpyval.Weibull is an instance of Weibull_, surpyval.CoxPH of CoxPH_, and so on. Call the methods on the instance you import (Weibull.fit(x)), not on the class. The model a fit returns has its own entry – for example Parametric for every univariate parametric fit – and each fitter’s Returns section names it.

Fitters take data in a common form: event times x, censoring flags c (0 observed, 1 right-censored, -1 left-censored, 2 interval-censored), counts n and truncation t (or tl / tr). Conventions defines these exactly, and Data Wrangling Examples shows how to build them from other formats. Not every model supports every kind of censoring or truncation; where one does not, its fit docstring says so.

• Non-Parametric
  • Non-Parametric Class
  • Non-Parametric Estimators
  • Other Non-Parametric Functions
• Parametric
  • Parametric Class
  • Distribution Classes
  • Discrete Distribution Classes
  • Special Distributions
  • Flexible Parametric (Royston-Parmar)
  • Custom Distributions
  • Zero-Failure Analysis (Weibayes)
  • Mixture Modelling
• Regression Modelling
  • Semi-Parametric Models
  • Parametric Models
  • Correlated Observations
• Competing Risks
  • Non-Parametric (Aalen-Johansen)
  • Parametric
  • Regression
• Recurrent Event Models
  • Recurrent Event (Poisson Process) Models
  • Non-Parametric Models
  • Renewal Models
  • Competing Risks (Marked) Models
  • Recurrent Event Regression Models
  • Trend Tests and Diagnostics
• Degradation Analysis
  • Degradation Analysis Fitter
  • Degradation Model
  • Path Models
  • Stress-Dependent Path Parameters
  • Step-Stress: the Accelerated Clock
  • Stochastic Process Models
  • Destructive Degradation
• Multivariate Modelling
  • Copulas
  • Fitted Model
  • Data
• Machine Learning (beta)
  • Random Survival Forest
  • Survival Tree
  • Tree Nodes
• Comparison Tests and Validation Metrics
  • Group-comparison tests
  • Restricted mean survival time
  • Automatic distribution selection
  • Prediction-validation metrics
• Saving and Loading Models
  • Infinite and NaN values
• Utilities
  • Data Classes
  • Data Wrangling Utilities
  • Abstract Base Classes
• Datasets
  • load_bearing_failures()
  • load_bofors_steel()
  • load_boston_housing()
  • load_framingham()
  • load_g1_kaminskiy_krivtsov()
  • load_heart_transplants()
  • load_kidney()
  • load_lung()
  • load_meeker_lfp()
  • load_mettas_and_zhao()
  • load_pbc2()
  • load_rossi_static()
  • load_rossi_time_varying()
  • load_sae()
  • load_support2()
  • load_tires_data()