The SEVERITY Procedure

Functional Summary

Table 23.1 summarizes the statements and options that control the SEVERITY procedure.

Table 23.1: SEVERITY Functional Summary

Description

Statement

Option

Statements

   

Specifies BY-group processing

BY

 

Specifies the response variable to model along with censoring and truncation effects

LOSS

 

Specifies the weight variable

WEIGHT

 

Specifies the classification variables

CLASS

 

Specifies the regression effects to model

SCALEMODEL

 

Specifies distributions to fit

DIST

 

Specifies the library to write scoring functions to

OUTSCORELIB

 

Specifies optimization options

NLOPTIONS

 

Specifies programming statements that define an objective function

Programming statements

Input and Output Options

   

Specifies that the OUTEST= data set contain covariance estimates

PROC SEVERITY

COVOUT

Specifies the input data set

PROC SEVERITY

DATA=

Specifies the input data set for parameter estimates

PROC SEVERITY

INEST=

Specifies the input item store for parameter initialization

PROC SEVERITY

INSTORE=

Limits the length of effect names

PROC SEVERITY

NAMELEN=

Specifies the output data set for CDF estimates

PROC SEVERITY

OUTCDF=

Specifies the output data set for parameter estimates

PROC SEVERITY

OUTEST=

Specifies the output data set for model information

PROC SEVERITY

OUTMODELINFO=

Specifies the output data set for statistics of fit

PROC SEVERITY

OUTSTAT=

Specifies the output item store for context and estimation results

PROC SEVERITY

OUTSTORE=

Data Interpretation Options

   

Specifies left-censoring

LOSS

LEFTCENSORED=

Specifies left-truncation

LOSS

LEFTTRUNCATED=

Specifies the probability of observability

LOSS

PROBOBSERVED=

Specifies right-censoring

LOSS

RIGHTCENSORED=

Specifies right-truncation

LOSS

RIGHTTRUNCATED=

Model Estimation Options

   

Specifies the model selection criterion

PROC SEVERITY

CRITERION=

Specifies the method for computing mixture distribution

SCALEMODEL

DFMIXTURE=

Specifies initial values for model parameters

DIST

INIT=

Specifies the objective function symbol

PROC SEVERITY

OBJECTIVE=

Specifies the offset variable in the scale regression model

SCALEMODEL

OFFSET=

Specifies the denominator for computing covariance estimates

PROC SEVERITY

VARDEF=

Empirical Distribution Function (EDF) Estimation Options

   

Specifies the confidence level for reporting the confidence interval for EDF estimates

PROC SEVERITY

EDFALPHA=

Specifies the nonparametric method of CDF estimation

PROC SEVERITY

EMPIRICALCDF=

Specifies the sample to be used for computing the EDF estimates

PROC SEVERITY

INITSAMPLE

EMPIRICALCDF=MODIFIEDKM Options

   

Specifies the $\alpha $ value for the lower bound on risk set size

PROC SEVERITY

ALPHA=

Specifies the c value for the lower bound on risk set size

PROC SEVERITY

C=

Specifies the absolute lower bound on risk set size

PROC SEVERITY

RSLB=

EMPIRICALCDF=TURNBULL Options

   

Specifies that the final EDF estimates be maximum likelihood estimates

PROC SEVERITY

ENSUREMLE

Specifies the relative convergence criterion

PROC SEVERITY

EPS=

Specifies the maximum number of iterations

PROC SEVERITY

MAXITER=

Specifies the threshold below which an EDF estimate is deemed to be 0

PROC SEVERITY

ZEROPROB=

Scoring Function Generation Options

   

Specifies that scoring functions of all models be written to one package

OUTSCORELIB

COMMONPACKAGE

Specifies the output data set for BY-group identifiers

OUTSCORELIB

OUTBYID=

Specifies the output library for scoring functions

OUTSCORELIB

OUTLIB=

Displayed Output and Plotting Options

   

Specifies that distributions be listed to the log without estimating any models that use them

DIST

LISTONLY

Limits or suppresses the display of class levels

PROC SEVERITY

NOCLPRINT

Suppresses all displayed and graphical output

PROC SEVERITY

NOPRINT

Specifies which graphical output to prepare

PROC SEVERITY

PLOTS=

Specifies which output to display

PROC SEVERITY

PRINT=

Specifies that distributions be validated without estimating any models that use them

DIST

VALIDATEONLY