What’s New in SAS/ETS 9.3
Overview
This chapter summarizes
the new features available in
SAS/ETS 9.3.
If you have used
SAS/ETS
procedures in the past, you can review this chapter to learn about
the new features that have been added. When you see a new feature
that might be useful for your work, see the appropriate chapter in
the
SAS/ETS User's Guide to read about the
feature in detail.
Highlights of Changes and Enhancements
The following new procedures
have been added to
SAS/ETS software:
-
COPULA procedure (experimental)
-
SSM procedure (experimental)
-
SASEXCCM interface engine (experimental)
New features have been
added to the following
SAS/ETS components:
-
-
-
-
SASEFAME interface engine
-
SASEHAVR interface engine
-
SASECRSP interface engine
-
-
-
The
SAS/ETS Model Editor
application, provided with
SAS/ETS 9.22 as an experimental interactive
graphical user interface for the MODEL procedure, is deprecated and
no longer documented in the
SAS/ETS User's Guide.
Highlights of Enhancements in SAS/ETS 9.22
AUTOREG Procedure
The AUTOREG procedure
now supports heteroscedasticity consistent covariance matrix estimators
(HCCME), which consistently estimate the covariance matrix even when
the heteroscedasticity structure might be unknown or misspecified.
Five forms of HCCMEs are supported: the plain sandwich form (HC0),
the degrees-of-freedom-adjustment form (HC1), two types of leverage-adjustment
forms (HC2 and HC3), and the high-leverage-adjustment form (HC4).
COPULA Procedure (Experimental)
The new experimental
COPULA procedure enables you to simulate realizations or estimate
parameters of multivariate distributions by using the copula approach.
This approach is based on the fact that a typical multivariate distribution
contains information about both the marginal behavior of individual
random variables and also about the dependence structure between them.
The COPULA procedure enables you to decouple these two effects and
model the dependence structure of random variables by linking their
cumulative distribution function (CDF) to a vector of their marginal
CDFs as described by the Sklar’s Theorem.
The COPULA procedure
supports the following types of distributions:
The COPULA procedure
can both estimate the parameters of copula models from data by using
maximum likelihood and simulate random data from copula distributions
by using either estimated or specified model parameters. The FIT statement
is used for model estimation, and the SIMULATE statement is used for
simulation. The PLOTS option in the FIT or SIMULATE statement provides
various ODS Graphics plots that help you analyze the underlying data.
ESM Procedure
New ODS plots and plot
options are available for the ESM procedure. You can plot the periodogram
for the error series or a combined pediodogram and spectral density
estimate plot.
SAS/ETS Model Editor Application (Experimental)
An experimental version
of a new interactive application, the
SAS/ETS Model Editor, was introduced
with
SAS/ETS 9.22. The
SAS/ETS Model Editor enables you to use the
powerful features of PROC MODEL through an interactive graphical user
interface.
Based on experience
with the experimental version in
SAS/ETS 9.22, plans for GUI features
to enable easier use of the MODEL procedure are being reconsidered.
This experimental
SAS/ETS Model Editor application is still available
with
SAS/ETS 9.3. However, because design changes are anticipated,
documentation for this application is not included in the
SAS/ETS
9.3 User’s Guide. Please refer to the
SAS/ETS
9.22 User’s Guide if you want to use the experimental
version of the
SAS/ETS Model Editor.
PANEL Procedure
The heteroscedasticity
consistent covariance matrix estimator (HCCME) was enhanced by adding
the CLUSTER option for the plain sandwich form (HC0), the degrees-of-freedom-adjusted
form (HC1), and two types of leverage-adjusted estimators (HC2 and
HC3). The CLUSTER option enables you to calculate a cluster-corrected
covariance matrix and provides cluster-adjusted standard errors for
parameter estimates.
SASECRSP Engine
The SASECRSP interface
engine enables you to access and process time series, events, portfolios,
and group data that reside in CRSPAccess (2.99 and earlier) legacy
databases.
It also provides a seamless
interface between CRSP, COMPUSTAT, and SAS data processing. Currently,
SASECRSP supports access of CRSP Stock databases, CRSP Indices databases,
and CRSP/Compustat Merged databases.
The following enhancement
has been made to the SASECRSP access engine:
-
Support has been added for Solaris
(SUNOS5.10).
SASEFAME Engine
The SASEFAME interface
engine provides a seamless interface between Fame and SAS data to
enable SAS users to access and process time series, case series, and
formulas that reside in a Fame database. The following enhancements
have been made to the SASEFAME access engine for Fame databases:
-
Support has been added for 64-bit
Windows.
-
Support has been added for AIX.
-
The SASEFAME interface uses FAME
10.
SASEHAVR Engine
The SASEHAVR interface
engine is a seamless interface between Haver and SAS data processing
that enables you to read economic and financial time series data that
reside in a Haver Analytics DLX (Data Link Express) database. The
following enhancements have been made to the SASEHAVR access engine
for Haver Analytics databases:
-
Support has been added for 64-bit
Windows.
SASEXCCM Engine (Experimental)
The new experimental
SASEXCCM interface engine enables you to access the CRSP/Compustat
Merged Database (CCM), created from data delivered via Compustat’s
Xpressfeed product, and the CRSP Stock, Indices, and Treasury Databases.
SASEXCCM provides a seamless interface for CRSP, Compustat, and SAS
data processing. The following new features are provided by the SASEXCCM
interface:
-
SETID= option supports item handling
data access to CRSPAccess (300 and up) databases with the designated
set identifier, including setid=250.
-
PERMNO= option enables you to select
based on permno, a primary keytype for CRSP Stock data.
-
PERMCO= option enables you to select
based on permco, a keytype for CRSP data.
-
CUSIP= option enables you to select
based on cusip, a keytype for CRSP data.
-
HCUSIP= option enables you to select
based on historical cusip, a keytype for CRSP data.
-
SICCD= option enables you to select
based on siccd, a keytype for CRSP data.
-
TICKER= option enables you to select
based on ticker, a keytype for CRSP data.
-
GVKEY= option enables you to select
based on gvkey, a primary keytype for COMPUSTAT data.
-
INDNO= option enables you to select
based on indno, a primary keytype for CRSP Indices data.
-
ITEMLIST= option specifies the
data items to be selected for access. This option accepts a string
in standard CRSP notation.
SEVERITY Procedure
The SEVERITY procedure
was experimental in
SAS/ETS 9.22. PROC SEVERITY is now production
status. The following new features and updates have been added to
the SEVERITY procedure:
-
The following updates have been
made to the syntax:
-
The MODEL statement is now replaced
with LOSS and SCALEMODEL statements. The LOSS statement specifies
the response variable along with any censoring and truncation information.
The SCALEMODEL statement specifies the regressor variables. The model-fitting
options that were specified in the MODEL statement in
SAS/ETS 9.22
should now be specified in the PROC SEVERITY statement.
-
You can now specify multiple distributions
in one DIST statement. You can also use a keyword to specify a group
of distributions. The syntax for specifying initial parameter values
of a distribution has also been updated. If you do not specify a DIST
statement, then PROC SEVERITY produces only the empirical CDF estimates
and does not fit all predefined distributions by default.
-
You can specify the number of occurrences
for each observation by using the new FREQ statement.
-
You can specify left-censoring
and right-truncation by using the new LEFTCENSORED= and RIGHTTRUNCATED=
options in the LOSS statement.
The method of specifying
censoring has been updated. Instead of using the indicator variable,
you now specify censoring by using a variable that contains the censoring
limit. This enables you to specify interval-censored data; that is,
data in which observations are both right-censored and left-censored.
For interval-censored
data, PROC SEVERITY uses Turnbull’s method to estimate the
empirical distribution function (EDF). Implementation of Turnbull’s
EDF estimation method is an experimental feature in
SAS/ETS 9.3.
-
Two predefined versions of Tweedie
distributions, TWEEDIE and STWEEDIE, can be fitted with PROC SEVERITY.
The TWEEDIE distribution has the more popular parameterization with
mean, dispersion, and index parameters. The STWEEDIE distribution
has an alternative parameterization with scale, Poisson mean, and
index parameters. The STWEEDIE distribution can be used for analyzing
regression effects.
-
You can estimate parameters by
minimizing your own objective function, which can be specified using
SAS programming statements. You can use various keyword functions
in your SAS program, which are internally expanded by PROC SEVERITY
with distribution-specific or problem-specific versions.
Note: This is an experimental feature
in
SAS/ETS 9.3.
-
You can now compute quantile and
limited moment for any distribution fitted with PROC SEVERITY by using
the two new functions, INVCDF and LIMMOMENT, respectively. These functions
are accessible in a PROC FCMP step.
SSM Procedure (Experimental)
The new experimental
SSM procedure enables linear state space modeling of time series and
longitudinal data. An important feature of the SSM procedure is a
modeling language that permits easy specification of possibly complex
state space models. In particular, the system matrices, such as the
state transition matrix and the covariance of the state disturbance,
can be time-varying and their elements can depend on user-specified
parameters in a complex way. Often a state space model can be specified
by combining simpler submodels. This modeling language is especially
suited for specification of such models. The following list identifies
the key features of the SSM procedure:
-
Many commonly needed state space
models, such as the basic univariate and multivariate structural time
series models, can be easily specified using a few keywords. Similarly,
models for panel data can also be easily specified.
-
The unknown model parameters are
estimated by (restricted) maximum likelihood and a variety of likelihood-based
information criteria are reported for model diagnostics.
-
One-step-ahead and full-sample
estimates of various state effects (linear combinations of the underlying
state vector) and one-step-ahead residuals can be output to a data
set. In particular, model-based forecasts, backcasts, interpolated
missing values of the response variables, and the estimates of the
latent effects such as trend, cycles, and seasonals, can be output
to a data set. These estimates are generated by using the Kalman filtering
and smoothing algorithm.
-
State space modeling is commonly
used for the analysis of regularly spaced univariate and multivariate
time series data. In fact, state space modeling is quite useful for
irregularly spaced, possibly with replicate measurements, longitudinal
data also. An important feature of the SSM procedure is that it enables
analysis of such longitudinal data, in addition to the regularly spaced
univariate and multivariate time series data. Several trend models
suitable for longitudinal data analysis can be easily specified using
a few keywords.
TCOUNTREG Procedure (Experimental)
The new experimental
TCOUNTREG procedure is an transitional version of the COUNTREG procedure.
It includes all features of the COUNTREG procedure. In addition to
features implemented in the COUNTREG procedure, PROC TCOUNTREG provides
the following new features:
-
Two new variable selection methods
are provided. The greedy search method can be used either with the
forward or backward selection. In each step, the AIC or BIC criterion
is evaluated, and the selection continues until the selection criterion
is met. The second method uses the penalized likelihood approach to
select significant variables. This method is not path-dependent as
in the case of greedy search, which falls into the family of LASSO
estimators. Using the penalized likelihood method, PROC TCOUNTREG
fits a model to the set of all candidate variables and evaluates it
simultaneously to find a subset of best-fitting variables.
-
Several conditional (fixed- and
random-effect) count panel data models have been added to the TCOUNTREG
procedure. The unconditional panel fixed-effect models can be easily
estimated in the TCOUNTREG procedure by using the CLASS statement
and the dummy variable approach. This technique is relatively simple
but is suitable only for a model with small number of cross sections.
If the number of cross sections is large, a conditional model is typically
preferred to overcome the incidental parameters problem. The TCOUNTREG
procedure enables you to estimate the following types of models:
-
Poisson regression model with fixed
and random effects
-
negative binomial regression model
with fixed and random effects
X12 Procedure
The following new features
have been added to the X12 procedure:
-
The PLOTS option in the PROC X12
statement now includes forecast plots. You can now request four different
plots for the forecast series on the original scale, and if the series
is transformed, on the transformed scale. The following values can
be specified in PLOTS=FORECAST(
value-list):
plots the actual time
series and its one-step-ahead forecasts over the historical period,
and plots the forecast and its confidence bands over the forecast
horizon.
plots the forecast
and its confidence bands over the forecast horizon only.
plots the one-step-ahead
model forecast and its confidence bands in the historical period.
plots the one-step-ahead
model forecast and its confidence bands in the historical period,
and plots the forecast and its confidence bands over the forecast
horizon.
plots the transformed
time series and its one-step-ahead forecast over the historical period,
and plots the forecast and its confidence bands over the forecast
horizon.
plots the forecast
of the transformed series and its confidence bands over the forecast
horizon only.
plots the one-step-ahead
model forecast of the transformed series and its confidence bands
in the historical period.
plots the one-step-ahead
model forecast of the transformed series and its confidence bands
in the historical period, and plots the forecast and its confidence
bands over the forecast horizon.
-
The following new values are available
in the PRINT= option in the AUTOMDL statement:
specifies that all
automatic modeling output be displayed.
suppresses all display
of automatic modeling output.
specifies that only
the requested automatic modeling tables be displayed.
-
The following new options are available
in the FORECAST statement:
specifies the number
of periods to backcast for regARIMA extension of the series. Backcasting
has been shown to improve seasonal adjustment for short series.
specifies that the
one-step-ahead forecasts be computed and displayed in addition to
the multistep forecasts. The one-step-ahead forecasts and associated
statistics are useful in evaluating the ARIMA model.
includes backcasts
in certain tables that are sent to the output data set.
includes forecasts
in certain tables that are sent to the output data set. This option
is an alias of the OUTFORECAST option in the X11 statement.
-
The FINAL=USER option in the X11
statement specifies that user-defined regressors are to be removed
from the final seasonally adjusted series.
-
The YEARSEAS option in the OUTPUT
statement specifies that variables containing values for year and
season are included in the OUT= data set. These values are useful
when creating seasonal plots.
-
An auxiliary variable has been
added to forecast data sets that are available through ODS OUTPUT.
The variable _SCALE_ indicates whether the observation refers to the
original series, "Original," or the transformed series, "Transformed."
The variable helps you subset the output when the series is transformed.
Copyright © SAS Institute Inc. All rights reserved.