funclp.Estimator module

class funclp.Estimator(distribution=None, **kwargs)

Bases: ABC, CudaReference

Class defining an estimator function for fitting data.

Parameters:

• distribution (Distribution) – Distribution instance.

• kwargs (dict) – Attributes to change.

deviance

How well the model fits the data

Type:

method

loss

Loss for gradient descent

Type:

method

observed

Observed Hessian (negative second derivative)

Type:

method

fisher

Expected Hessian (Fisher information)

Type:

method

abstract property cpukernel_deviance

abstract property cpukernel_fisher

abstract property cpukernel_loss

abstract property cpukernel_observed

abstractmethod deviance(raw_data, model_data, weights=1, **kwargs)

abstractmethod fisher(raw_data, model_data, weights=1, **kwargs)

abstract property gpukernel_deviance

abstract property gpukernel_fisher

abstract property gpukernel_loss

abstract property gpukernel_observed

abstractmethod loss(raw_data, model_data, weights=1, **kwargs)

property name

abstractmethod observed(raw_data, model_data, weights=1, **kwargs)

Estimators

• funclp.LSE module
• funclp.MLE module