abs_inv

pred

t_val

A loss function for survival data. Equal to 
  
  <code>mean(abs(1/(pred+1) - 1/(t_val+1 ) ) )</code>. 
  
  This function heavily penalizes inaccurate predictions that are closer to the 
  origin than those that are farther way. In otherwords, it will favor a model
  that accurately predicts high risk subjects over a model that accurately predicts
  loss risk subjects.

Regression models for interval censored data. Currently supports Cox-PH,  proportional odds, and accelerated failure time models. Allows for both semi and fully parametric models (parametric only for accelerated failure time models). Includes functions for easy visual diagnostics of model fits and imputation of censored data.

abs_inv: Loss Function for interval censored cross validation

Description

Usage

Arguments