sparsesurv.loss module

Summary

Functions:

aft_negative_likelihood	Negative log-likelihood function for accelerated failure time model.
aft_negative_likelihood_beta	Negative log-likelihood function for accelerated failure time model when the design matrix
breslow_negative_likelihood	Negative log-likelihood function with Breslow tie-correction.
breslow_negative_likelihood_beta
efron_negative_likelihood	Negative log-likelihood function with Efron tie-correction.
efron_negative_likelihood_beta	Negative log-likelihood function with Efron tie-correction when the design matrix
eh_negative_likelihood	Negative log-likelihood function for extended hazards model.
eh_negative_likelihood_beta	Negative log-likelihood function for extended hazards model when the design matrix

Reference

breslow_negative_likelihood(linear_predictor, time, event)

Negative log-likelihood function with Breslow tie-correction.

Parameters:

• linear_predictor (npt.NDArray[np.float64]) – Linear predictor of the training data.

• time (npt.NDArray[np.float64]) – Time of the training data of length n. Assumed to be sorted.

• event (npt.NDArray[np.float64]) – Event indicator of the training data of length n.

Raises:

RuntimeError – Raises runtime error when there are no deaths/events in the given batch of samples.

Returns:

Average negative log likelihood.

Return type:

npt.NDArray[np.float64]

breslow_negative_likelihood_beta(beta, X, time, event)

efron_negative_likelihood(linear_predictor, time, event)

Negative log-likelihood function with Efron tie-correction.

Parameters:

• linear_predictor (npt.NDArray[np.float64]) – Linear predictor of the training data.

• time (npt.NDArray[np.float64]) – Time of the training data of length n. Assumed to be sorted.

• event (npt.NDArray[np.float64]) – Event indicator of the training data of length n.

Raises:

RuntimeError – Raises runtime error when there are no deaths/events in the given batch of samples.

Returns:

Average negative log likelihood.

Return type:

npt.NDArray[np.float64]

efron_negative_likelihood_beta(beta, X, time, event)

Negative log-likelihood function with Efron tie-correction when the design matrix

and the coefficients beta are provided instead of the linear predictor directly.

Parameters:

• beta (npt.NDArray[np.float64]) – Coefficient vector of length p.

• X (npt.NDArray[np.float64]) – Design matrix of the training data. N rows and p columns.

• time (npt.NDArray[np.float64]) – Time of the training data of length n. Assumed to be sorted.

• event (npt.NDArray[np.float64]) – Event indicator of the training data of length n.

Returns:

Average negative log likelihood.

Return type:

npt.NDArray[np.float64]

aft_negative_likelihood(linear_predictor, time, event, bandwidth=None)

Negative log-likelihood function for accelerated failure time model.

Parameters:

• linear_predictor (npt.NDArray[np.float64]) – Linear predictor of the training data.

• time (npt.NDArray[np.float64]) – Time of the training data of length n. Assumed to be sorted.

• event (npt.NDArray[np.float64]) – Event indicator of the training data of length n.

• bandwidth (float, optional) – Bandwidth to kernel-smooth the profile likelihood. Will be estimated empirically if not specified. Defaults to None.

Raises:

RuntimeError – Raises runtime error when there are no deaths/events in the given batch of samples.

Returns:

Average negative log-likelihood.

Return type:

npt.NDArray[np.float64]

aft_negative_likelihood_beta(beta, X, time, event, bandwidth=None)

Negative log-likelihood function for accelerated failure time model when the design matrix

and the coefficients beta are provided instead of the linear predictor directly.

Parameters:

• beta (npt.NDArray[np.float64]) – Coefficient vector of length p.

• X (npt.NDArray[np.float64]) – Design matrix of the training data. N rows and p columns.

• time (npt.NDArray[np.float64]) – Time of the training data of length n. Assumed to be sorted.

• event (npt.NDArray[np.float64]) – Event indicator of the training data of length n.

• bandwidth (float, optional) – Bandwidth to kernel-smooth the profile likelihood. Will be estimated empirically if not specified. Defaults to None.

Returns:

Average negative log-likelihood.

Return type:

npt.NDArray[np.float64]

eh_negative_likelihood(linear_predictor, time, event, bandwidth=None)

Negative log-likelihood function for extended hazards model.

Parameters:

• linear_predictor (npt.NDArray[np.float64]) – Linear predictor of the training data.

• time (npt.NDArray[np.float64]) – Time points of the training data of length n. Assumed to be sorted.

• event (npt.NDArray[np.float64]) – Event indicator of the training data of length n.

• bandwidth (float, optional) – Bandwidth to kernel-smooth the profile likelihood. Will be estimated empirically if not specified. Defaults to None.

Raises:

RuntimeError – Raises runtime error when there are no deaths/events in the given batch of samples.

Returns:

Average negative log-likelihood.

Return type:

npt.NDArray[np.float64]

eh_negative_likelihood_beta(beta, X, time, event, bandwidth=None)

Negative log-likelihood function for extended hazards model when the design matrix

and the coefficients beta are provided instead of the linear predictor directly.

Parameters:

• beta (npt.NDArray[np.float64]) – Coefficient vector of length p.

• X (npt.NDArray[np.float64]) – Design matrix of the training data. N rows and p columns.

• time (npt.NDArray[np.float64]) – Time of the training data of length n. Assumed to be sorted.

• event (npt.NDArray[np.float64]) – Event indicator of the training data of length n.

• bandwidth (float, optional) – Bandwidth to kernel-smooth the profile likelihood. Will be estimated empirically if not specified. Defaults to None.

Returns:

Average negative log-likelihood.

Return type:

npt.NDArray[np.float64]