risk_score: risk scores

In herglola/SVHM: Support-Vector-Hazard-Machine

risk scores

Description

calculates the risk scores for one individual with the help of the calculated optimal solution to the quadratic programming problem of SVHM and the kernel matrix of the covariates of the test dataset.

Usage

risk_score(
  gamma_sol,
  event_vec,
  v,
  covariates_train,
  num_event_times,
  gamma_squared = 0.5,
  d = 1
)

Arguments

gamma_sol	optimal solution of the SVHM
event_vec	vector containing information of the training if a subject in the training dataset is at risk or if an event happens. If n are the number of subjects in the training dataset and m the number of event times in the training dataset, then event_vec has length n*m
v	covariates of the individual for which the risk is to be calculated
covariates_train	dataset of covariates of the subjects in the training dataset
num_event_times	number of event times that occour in the training data set
gamma_squared	width of gaussian kernel
d	degree of polynomial kernel
type	Type of kernel, either 'gauss' or 'poly' for gaussian or polynomial kernel

Value

risk score of the individual

Note

The calculated risk score is not the actual risk scores defined by the Risk function but it induce an ordering of the risk scores. For detailed information see reference

References

Wang, Y., Chen, T., and Zeng, D. Support vector hazards machine: A counting process framework for learning risk scores for censored outcomes. Journal of Machine Learning Research, 17(167):1-37, 2016

herglola/SVHM documentation built on March 24, 2022, 12:44 p.m.