DOCUMENTATION
Sign In

Get access to My SAS, trials, communities and more.

Econometrics Procedures
 

HMM Procedure

Data Table Output

The HMM procedure can create the data tables that are specified in the OUT= suboption of the LABELSWITCH= option in the PROC HMM statement; the OUTSTAT= option in the PROC HMM statement; the OUT= options in the DECODE, EVALUATE, FILTER, and SMOOTH statements; the OUT= and OUTALL= options in the ESTIMATE and FORECAST statements; and the OUTMODEL= option in the SCORE statement. The column information for each table is described in the following sections.

OUT= Data Table Generated from the LABELSWITCH= Option in the PROC HMM Statement

The output data table contains the following variables:

ModelIndex

model index, if multiple models are estimated or scored

NState

number of states, if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

YLag

order of the autoregressive process, if you specify a range of orders of the autoregressive processes in the YLAG=n 1 colon n 2 option

OldStateLabel

old label of the state

OldComponentLabel

old label of the component for a GM HMM

NewStateLabel

new label of the state

NewComponentLabel

new label of the component for a GM HMM

OUTSTAT= Data Table Generated from the PROC HMM Statement

The output data table contains the following variables:

ModelIndex

model index, if multiple models are estimated or scored

NState

number of states, if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

YLag

order of the autoregressive process, if you specify a range of orders of the autoregressive processes in the YLAG=n 1 colon n 2 option

LogLikelihood

log likelihood

AIC

Akaike’s information criterion

AICC

corrected Akaike’s information criterion

BIC

Bayesian information criterion (also referred to as the Schwarz Bayesian criterion, SBC)

HQC

Hannan-Quinn criterion

OUT= Data Table Generated from the DECODE Statement

The output data table contains the following variables:

ModelIndex

model index, if multiple models are estimated or scored

NState

number of states, if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

YLag

order of the autoregressive process, if you specify a range of orders of the autoregressive processes in the YLAG=n 1 colon n 2 option

Section ID Variable

values of the variable that is specified in the SECTION= option in the ID statement

Time ID Variable

values of the variable that is specified in the TIME= option in the ID statement

State

state value in the most possible path

OUT= Data Table Generated from the ESTIMATE Statement

The output data table contains the following variables:

ModelIndex

model index, if multiple models are estimated or scored

NState

number of states, if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

YLag

order of the autoregressive process, if you specify a range of orders of the autoregressive processes in the YLAG=n 1 colon n 2 option

Index

index of the parameter

Type

type of values in the Statek variables: EST for estimate and STD for standard error

Parameter

parameter name

Statek

value for state k, k equals 1 comma ellipsis comma upper K, where K is the number of states if you specify the NSTATE=K option; otherwise, upper K equals max left-parenthesis n 1 comma n 2 right-parenthesis, the maximum of the range of numbers of states if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

OUTALL= Data Table Generated from the ESTIMATE Statement

The output data table contains the following variables:

ModelIndex

model index, if multiple models are estimated or scored

NState

number of states, if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

YLag

order of the autoregressive process, if you specify a range of orders of the autoregressive processes in the YLAG=n 1 colon n 2 option

Index

index of the parameter

Parameter

parameter name

Estimate

parameter estimate

StdErr

standard error

COV_n

value for column n, n equals 1 comma ellipsis comma upper N, in the covariance matrix of parameter estimates, where N is the number of parameters if only one model is estimated or scored; otherwise, N is the maximum of numbers of parameters for all models

OUT= Data Table Generated from the EVALUATE Statement

The output data table contains the following variables:

ModelIndex

model index, if multiple models are estimated or scored

NState

number of states, if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

YLag

order of the autoregressive process, if you specify a range of orders of the autoregressive processes in the YLAG=n 1 colon n 2 option

Section ID Variable

values of the variable that is specified in the SECTION= option in the ID statement

Time ID Variable

values of the variable that is specified in the TIME= option in the ID statement

LogLikelihood

log likelihood

OUT= Data Table Generated from the FILTER Statement

The output data table contains the following variables:

ModelIndex

model index, if multiple models are estimated or scored

NState

number of states, if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

YLag

order of the autoregressive process, if you specify a range of orders of the autoregressive processes in the YLAG=n 1 colon n 2 option

Section ID Variable

values of the variable that is specified in the SECTION= option in the ID statement

Time ID Variable

values of the variable that is specified in the TIME= option in the ID statement

Statek

filtered probability value for state k, k equals 1 comma ellipsis comma upper K, where K is the number of states if you specify the NSTATE=K option; otherwise, upper K equals max left-parenthesis n 1 comma n 2 right-parenthesis, the maximum of the range of numbers of states if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

OUT= Data Table Generated from the FORECAST Statement

The output data table contains the following variables:

ModelIndex

model index, if multiple models are estimated or scored

NState

number of states, if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

YLag

order of the autoregressive process, if you specify a range of orders of the autoregressive processes in the YLAG=n 1 colon n 2 option

Section ID Variable

values of the variable that is specified in the SECTION= option in the ID statement

Time ID Variable

values of the variable that is specified in the TIME= option in the ID statement

Step

value of h in the h-step-ahead forecast

Statek

forecasted probability value for state k, k equals 1 comma ellipsis comma upper K, where K is the number of states if you specify the NSTATE=K option; otherwise, upper K equals max left-parenthesis n 1 comma n 2 right-parenthesis, the maximum of the range of numbers of states if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

{DVp}_Forecast

mean forecast for dependent variable p, p equals 1 comma ellipsis comma k Subscript y Baseline, where k Subscript y is the number of dependent variables; {DVp} stands for the name of dependent variable p

Q1

first quantile value, left-parenthesis 1 minus alpha right-parenthesis slash 2, where alpha is specified in the ALPHA=alpha option in the FORECAST statement

{DVp}_Q1

first quantile forecast for dependent variable p, p equals 1 comma ellipsis comma k Subscript y Baseline, where k Subscript y is the number of dependent variables; {DVp} stands for the name of dependent variable p

Q2

second quantile value, left-parenthesis 1 plus alpha right-parenthesis slash 2, where alpha is specified in the ALPHA=alpha option in the FORECAST statement

{DVp}_Q2

second quantile forecast for dependent variable p, p equals 1 comma ellipsis comma k Subscript y Baseline, where k Subscript y is the number of dependent variables; {DVp} stands for the name of dependent variable p

Q3

third quantile value, 1 slash 2, for median

{DVp}_Q3

median forecast for dependent variable p, p equals 1 comma ellipsis comma k Subscript y Baseline, where k Subscript y is the number of dependent variables; {DVp} stands for the name of dependent variable p

OUTALL= Data Table Generated from the FORECAST Statement

The output data table contains the following variables:

ModelIndex

model index, if multiple models are estimated or scored

NState

number of states, if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

YLag

order of the autoregressive process, if you specify a range of orders of the autoregressive processes in the YLAG=n 1 colon n 2 option

Section ID Variable

values of the variable that is specified in the SECTION= option in the ID statement

Time ID Variable

values of the variable that is specified in the TIME= option in the ID statement

Step

value of h in the h-step-ahead forecast

Index

index of the dependent variable

Variable

dependent variable name

Statek

forecasted probability value for state k, k equals 1 comma ellipsis comma upper K, where K is the number of states if you specify the NSTATE=K option; otherwise, upper K equals max left-parenthesis n 1 comma n 2 right-parenthesis, the maximum of the range of numbers of states if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

Forecast

mean forecast for the dependent variable

StdErr

standard error of forecast for the dependent variable

LowerCL

lower confidence limit of forecast for the dependent variable

UpperCL

upper confidence limit of forecast for the dependent variable

Median

median forecast for the dependent variable

COV_{DVp}

value for column p, p equals 1 comma ellipsis comma k Subscript y Baseline, in the covariance matrix of forecasts of dependent variables, where k Subscript y is the number of dependent variables; {DVp} stands for the name of the dependent variable p

OUTMODEL= Data Table Generated from the SCORE Statement

The output data table consists of binary large object columns to store the binary data, which should be consumed only by the INMODEL= option in the SCORE statement. The number of columns and the definition of columns might be changed without notice in future releases. Do not edit the content of this output data table. If you run the PRINT procedure on this data table, you get a row of zeros or meaningless characters because the data type of each column is binary.

OUT= Data Table Generated from the SMOOTH Statement

The output data table contains the following variables:

ModelIndex

model index, if multiple models are estimated or scored

NState

number of states, if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

YLag

order of the autoregressive process, if you specify a range of orders of the autoregressive processes in the YLAG=n 1 colon n 2 option

Section ID Variable

values of the variable that is specified in the SECTION= option in the ID statement

Time ID Variable

values of the variable that is specified in the TIME= option in the ID statement

Statek

smoothed probability value for state k, k equals 1 comma ellipsis comma upper K, where K is the number of states if you specify the NSTATE=K option; otherwise, upper K equals max left-parenthesis n 1 comma n 2 right-parenthesis, the maximum of the range of numbers of states if you specify a range of numbers of states in the NSTATE=n 1 colon n 2 option

Last updated: July 09, 2026