gbnet.models.survival.hazard_integrator

Classes

HazardIntegrator

Functions

loadModule(module)

Load the appropriate gradient boosting module.

Module Contents

gbnet.models.survival.hazard_integrator.loadModule(module)

Load the appropriate gradient boosting module.

class gbnet.models.survival.hazard_integrator.HazardIntegrator(covariate_cols=[], params={}, min_hess=0.0, module_type='XGBModule', integration_method='trapezoid')

Bases: torch.nn.Module

Parameters:

• covariate_cols (List[str])

• params (Dict)

• min_hess (float)

• module_type (str)

• integration_method (str)

params

min_hess = 0.0

module_type = 'XGBModule'

integration_method = 'trapezoid'

covariate_cols = ['time']

gb_module: object | None = None

Module

static_data: Dict[str, torch.Tensor]

_integrate_slice(values, dt, same_unit)

_prepare_data(df)

Pre-processes and caches data that is static during training. This method performs sorting, tensor conversion, and computes time differences and group boundaries once.

Parameters:

df (pandas.DataFrame)

forward(df, return_survival_estimates=True)

Parameters:

• df (pandas.DataFrame)

• return_survival_estimates (bool)

Return type:

Dict[str, Any]

gb_step()

Triggers the gradient boosting model to take a step.