statsmodels.duration.hazard_regression.PHReg

class statsmodels.duration.hazard_regression.PHReg(endog, exog, status=None, entry=None, strata=None, offset=None, ties='breslow', missing='drop', **kwargs)

Cox Proportional Hazards Regression Model

The Cox PH Model is for right censored data.

Parameters:

endogarray_like

The observed times (event or censoring)

exog2D array_like

The covariates or exogeneous variables

statusarray_like

The censoring status values; status=1 indicates that an event occurred (e.g. failure or death), status=0 indicates that the observation was right censored. If None, defaults to status=1 for all cases.

entryarray_like

The entry times, if left truncation occurs

strataarray_like

Stratum labels. If None, all observations are taken to be in a single stratum.

tiesstr

The method used to handle tied times, must be either ‘breslow’ or ‘efron’.

offsetarray_like

Array of offset values

missingstr

The method used to handle missing data

Attributes:

endog_names

Names of endogenous variables

exog_names

Names of exogenous variables

Methods

baseline_cumulative_hazard(params)	Estimate the baseline cumulative hazard and survival functions
baseline_cumulative_hazard_function(params)	Returns a function that calculates the baseline cumulative hazard function for each stratum
breslow_gradient(params)	Returns the gradient of the log partial likelihood, using the Breslow method to handle tied times
breslow_hessian(params)	Returns the Hessian of the log partial likelihood evaluated at params, using the Breslow method to handle tied times
breslow_loglike(params)	Returns the value of the log partial likelihood function evaluated at params, using the Breslow method to handle tied times
efron_gradient(params)	Returns the gradient of the log partial likelihood evaluated at params, using the Efron method to handle tied times
efron_hessian(params)	Returns the Hessian matrix of the partial log-likelihood evaluated at params, using the Efron method to handle tied times
efron_loglike(params)	Returns the value of the log partial likelihood function evaluated at params, using the Efron method to handle tied times
fit([groups])	Fit a proportional hazards regression model
fit_regularized([method, alpha, ...])	Return a regularized fit to a proportional hazards regression model
from_formula(formula, data[, status, entry, ...])	Create a proportional hazards regression model from a formula and dataframe
get_distribution(params[, scale, exog])	Returns a scipy distribution object corresponding to the distribution of uncensored endog (duration) values for each case
hessian(params)	Returns the Hessian matrix of the log partial likelihood function evaluated at params
information(params)	Fisher information matrix of model
initialize()	Initialize (possibly re-initialize) a Model instance
loglike(params)	Returns the log partial likelihood function evaluated at params
predict(params[, exog, cov_params, endog, ...])	Returns predicted values from the proportional hazards regression model
robust_covariance(params)	Returns a covariance matrix for the proportional hazards model regression coefficient estimates that is robust to certain forms of model misspecification
score(params)	Returns the score function evaluated at params
score_residuals(params)	Returns the score residuals calculated at a given vector of parameters
weighted_covariate_averages(params)	Returns the hazard-weighted average of covariate values for subjects who are at-risk at a particular time

Notes

Proportional hazards regression models should not include an explicit or implicit intercept. The effect of an intercept is not identified using the partial likelihood approach.

endog, event, strata, entry, and the first dimension of exog all must have the same length

Methods

baseline_cumulative_hazard(params)	Estimate the baseline cumulative hazard and survival functions
baseline_cumulative_hazard_function(params)	Returns a function that calculates the baseline cumulative hazard function for each stratum
breslow_gradient(params)	Returns the gradient of the log partial likelihood, using the Breslow method to handle tied times
breslow_hessian(params)	Returns the Hessian of the log partial likelihood evaluated at params, using the Breslow method to handle tied times
breslow_loglike(params)	Returns the value of the log partial likelihood function evaluated at params, using the Breslow method to handle tied times
efron_gradient(params)	Returns the gradient of the log partial likelihood evaluated at params, using the Efron method to handle tied times
efron_hessian(params)	Returns the Hessian matrix of the partial log-likelihood evaluated at params, using the Efron method to handle tied times
efron_loglike(params)	Returns the value of the log partial likelihood function evaluated at params, using the Efron method to handle tied times
fit([groups])	Fit a proportional hazards regression model
fit_regularized([method, alpha, ...])	Return a regularized fit to a proportional hazards regression model
from_formula(formula, data[, status, entry, ...])	Create a proportional hazards regression model from a formula and dataframe
get_distribution(params[, scale, exog])	Returns a scipy distribution object corresponding to the distribution of uncensored endog (duration) values for each case
hessian(params)	Returns the Hessian matrix of the log partial likelihood function evaluated at params
information(params)	Fisher information matrix of model
initialize()	Initialize (possibly re-initialize) a Model instance
loglike(params)	Returns the log partial likelihood function evaluated at params
predict(params[, exog, cov_params, endog, ...])	Returns predicted values from the proportional hazards regression model
robust_covariance(params)	Returns a covariance matrix for the proportional hazards model regression coefficient estimates that is robust to certain forms of model misspecification
score(params)	Returns the score function evaluated at params
score_residuals(params)	Returns the score residuals calculated at a given vector of parameters
weighted_covariate_averages(params)	Returns the hazard-weighted average of covariate values for subjects who are at-risk at a particular time

Properties

endog_names	Names of endogenous variables
exog_names	Names of exogenous variables