The HPFMM Procedure

ODS Table Names

Each table created by PROC HPFMM has a name associated with it, and you must use this name to reference the table when you use ODS statements. These names are listed in Table 6.10.

Table 6.10: ODS Tables Produced by PROC HPFMM

Table Name

Description

Required Statement / Option

Autocorr

Autocorrelation among posterior estimates

BAYES

BayesInfo

Basic information about Bayesian estimation

BAYES

ClassLevels

Level information from the CLASS statement

CLASS

CompDescription

Component description in models with varying number of components

KMAX= in MODEL with ML estimation

CompEvaluation

Comparison of mixture models with varying number of components

KMAX= in MODEL with ML estimation

CompInfo

Component information

COMPONENTINFO option in PROC HPFMM statement

ConvergenceStatus

Status of optimization at conclusion of optimization

Default output

Constraints

Linear equality and inequality constraints

RESTRICT statement or EQUATE=EFFECTS option in MODEL statement

Corr

Asymptotic correlation matrix of parameter estimates (ML) or empirical correlation matrix of the Bayesian posterior estimates

CORR option in PROC HPFMM statement

Cov

Asymptotic covariance matrix of parameter estimates (ML) or empirical covariance matrix of the Bayesian posterior estimates

COV option in PROC HPFMM statement

CovI

Inverse of the covariance matrix of the parameter estimates

COVI option in PROC HPFMM statement

ESS

Effective sample size

DIAG=ESS option in BAYES statement

FitStatistics

Fit statistics

Default output

Geweke

Geweke diagnostics (Geweke, 1992) for Markov chain

DIAG=GEWEKE option in BAYES statement

Hessian

Hessian matrix from the maximum likelihood optimization, evaluated at the converged estimates

HESSIAN

IterHistory

Iteration history

Default output for ML estimation

MCSE

Monte Carlo standard errors

DIAG=MCERROR in BAYES statement

MixingProbs

Solutions for the parameter estimates associated with effects in PROBMODEL statements

Default output for ML estimation if number of components is greater than 1

ModelInfo

Model information

Default output

NObs

Number of observations read and used, number of trials and events

Default output

OptInfo

Optimization information

Default output for ML estimation

ParameterEstimates

Solutions for the parameter estimates associated with effects in MODEL statements

Default output for ML estimation

ParameterMap

Mapping of parameter names to OUTPOST= data set

OUTPOST= option in BAYES statement

PriorInfo

Prior distributions and initial value of Markov chain

BAYES

PostSummaries

Summary statistics for posterior estimates

BAYES

PostIntervals

Equal-tail and highest posterior density intervals for posterior estimates

BAYES

ResponseProfile

Response categories and category modeled

Default output in models with binary response