The following features have been added to the SEVERITY procedure:
The SEVERITY procedure now supports a new OFFSET= option in the SCALEMODEL statement to specify an offset variable in the scale regression model. An offset variable is a regressor that has a fixed regression coefficient of 1. An offset variable is useful for modeling the scale parameter per unit of some measure of exposure and for correcting the omitted variable bias.
The SEVERITY procedure now supports a new OUTSCORELIB statement to create scoring functions. A scoring function is derived from a distribution function such that the parameter estimates of a fitted severity model are encoded inside the scoring function. You can use a scoring function to evaluate the corresponding distribution function without having to write a potentially complex SAS program to extract the parameter estimates. If you have specified a scale regression model, then you can use the scoring functions to score new observations by easily computing various statistics of interest such as the mean and quantiles that depend on the regressor data. Note: This is an experimental feature.
The following two features, which were experimental in previous releases, are now production:
specification of custom objective functions
estimation of the empirical distribution function (EDF) with Turnbull’s estimator when you specify left-censoring or right-truncation in addition to right-censoring or left-truncation