PROC HPFENGINE options ;

The following options can be used in the PROC HPFENGINE statement.

BACK=n

specifies the number of observations before the end of the data where the multistep forecasts are to begin. This option is often used to obtain performance statistics. See the PRINT= option details about printing performance statistics. The default is BACK=0.

DATA=SAS-data-set

names the SAS data set that contains the input data for the procedure to forecast. If the DATA= option is not specified, the most recently created SAS data set is used.

ERRORCONTROL= ( SEVERITY= ( severity-options ) STAGE= ( stage-options ) MAXMESSAGE=number )

enables finer control of message printing. The error severity level and HPFENGINE procedure processing stages are set independently. A logical 'and' is taken over all the specified options, and any message that tests true against the results of the 'and' is printed.

Available severity-options are as follows:

LOW

specifies that only low severity, minor issues be printed.

MEDIUM

specifies that only medium-severity problems be printed.

HIGH

specifies that only severe errors be printed.

ALL

specifies that errors of all severity levels (LOW, MEDIUM, and HIGH) be printed.

NONE

specifies that no messages from the HPFENGINE procedure be printed.

Available stage-options are as follows:

PROCEDURELEVEL

specifies that only errors that occur during option processing and validation be printed.

DATAPREP

specifies that only errors that occur during the accumulation of data and the application of SETMISS= and ZEROMISS= options be printed.

SELECTION

specifies that only errors that occur during the model selection process be printed.

ESTIMATION

specifies that only errors that occur during the model parameter estimation process be printed.

FORECASTING

specifies that only errors that occur during the model evaluation and forecasting process be printed.

ALL

is the same as specifying all PROCEDURELEVEL, DATAPREP, SELECTION, ESTIMATION, and FORECASTING options.

Examples are as follows:

   errorcontrol=(severity=(high medium) stage=all);

prints high- and moderate-severity errors at any processing stage of PROC HPFENGINE.

   errorcontrol=(severity=high stage=dataprep);

prints high-severity errors only during the data preparation.

   errorcontrol=(severity=none stage=all);

turns off messages from PROC HPFENGINE.

   errorcontrol=( severity=(high medium low)
                  stage=( procedurelevel dataprep selection
                          estimation forecasting) );

specifies the default behavior. Also the following statement specifies the default behavior:

   errorcontrol=(severity=all stage=all)

EXCEPTIONS=except-option

specifies the desired handling of arithmetic exceptions during the run. You can specify except-option as one of the following:

IGNORE

specifies that PROC HPFENGINE stop on an arithmetic exception. No recovery is attempted. This is the default behavior if the EXCEPTIONS= option is not specified.

CATCH(ESM)

specifies that PROC HPFENGINE generate a forecast based on using its default exponential smoothing model for the variable that produces the arithmetic exception in the current BY group. The ESM model is equivalent to the default model used by the HPFESMSPEC procedure with no modifications.

CATCH(RW)

specifies that PROC HPFENGINE generate a forecast based on a zero-drift random walk model for the variable that produces the exception in the current BY group.

CATCH(MISSING)

specifies that PROC HPFENGINE generate a forecast of missing values for the variable that produces the exception in the current BY group.

The order of CATCH handling corresponds to the order of the preceding list. If CATCH(ESM) handling produces an arithmetic exception, it attempts to generate a forecast by using CATCH(RW) semantics. Likewise, if CATCH(RW) handling produces an arithmetic exception, it generates a missing value forecast.

GLOBALSELECTION=catalog-name

specifies the name of a catalog entry that serves as a model selection list. This is the selection list used to forecast all series if no INEST= data set is provided. It is also the selection list used if individual model selections are missing in the INEST= data set when INEST= is provided. If REPOSITORY= is not present, GLOBALSELECTION defaults to BEST, specified in SASHELP.HPFDFLT.

IGNORECHOOSE

specifies that the CHOOSE= option in the HPFSELECT procedure be ignored when selecting a model in the candidate model list. The best model is selected regardless of the model chosen in the CHOOSE= option in the HPFSELECT procedure.

INEST=SAS-data-set

contains information that maps forecast variables to models or selection lists, and data set variables to model variables. It can also contain parameter estimates used if the TASK=FORECAST or TASK=UPDATE options are present. INEST= is optional. See the description of the GLOBALSELECTION= option for more information.

INEVENT=SAS-data-set

contains information that describes predefined events. This data set is usually created by the HPFEVENTS procedure. This option is only used if events are included in a model.

LEAD=n

specifies the number of periods ahead to forecast (forecast lead or horizon). The default is LEAD=12.

The LEAD= value is relative to the last observation in the input data set and not to the last nonmissing observation of a particular series. Thus, if a series has missing values at the end, the actual number of forecasts computed for that series will be greater than the LEAD= value.

OUT=SAS-data-set

names the output data set to contain the forecasts of the variables specified in the subsequent FORECAST statements. If an ID variable is specified, it will also be included in the OUT= data set. The values are accumulated based on the ACCUMULATE= option and forecasts are appended to these values. If the OUT= data set is not specified, a default output data set DATAn is created. If you do not want the OUT= data set created, then use OUT=_NULL_.

OUTCOMPONENT=SAS-data-set

names the output data set to contain the forecast components. The components included in the output depend on the model.

OUTEST=SAS-data-set

contains information that maps forecast variables to model specifications, and data set variables to model variables and parameter estimates.

An OUTEST= data set will frequently be used as the INEST= data set for subsequent invocations of PROC HPFENGINE. In such a case, if the PROC HPFENGINE statement option TASK=FORECAST is used, forecasts are generated using the parameter estimates found in this data set and are not reestimated.

OUTFOR=SAS-data-set

names the output data set to contain the forecast time series components (actual, predicted, lower confidence limit, upper confidence limit, prediction error, and prediction standard error). The OUTFOR= data set is particularly useful for displaying the forecasts in tabular or graphical form.

OUTINDEP=SAS-data-set

names the output data set to contain input in the forecasting process. This information is useful if future values of input variables are automatically supplied by the HPFENGINE procedure. Such a case would occur if one or more input variables are listed in either the STOCHASTIC or the CONTROLLABLE statement and if there are missing future values of these input variables.

OUTMODELINFO=SAS-data-set

names the output data set to contain detailed information about the selected forecast model. The data set has information such as the model family, presence or absence of inputs, events and outliers, and so forth.

OUTPROCINFO=SAS-data-set

names the output data set to contain information in the SAS log, specifically the number of notes, errors, and warnings and the number of series processed, forecasts requested, and forecasts failed.

OUTSTAT=SAS-data-set

names the output data set to contain the statistics of fit (or goodness-of-fit statistics). The OUTSTAT= data set is useful for evaluating how well the model fits the series. The statistics of fit are based on the entire range of the time series.

OUTSTATSELECT=SAS-data-set

names the output data set to contain statistics of fit for all of the candidate models fit during model selection. The OUTSTATSELECT= data set is useful for comparing the performance of various models.

PLOT=option | (options )

specifies the graphical output desired. By default, the HPFENGINE procedure produces no graphical output. The following printing options are available:

ERRORS

plots prediction error time series graphics.

ACF

plots prediction error autocorrelation function graphics.

PACF

plots prediction error partial autocorrelation function graphics.

IACF

plots prediction error inverse autocorrelation function graphics.

WN

plots white noise graphics.

FORECASTS

plots forecast graphics.

FORECASTCYCLES

plots forecast seasonal cycles graphics.

FORECASTSONLY

plots the forecast in the forecast horizon only.

COMPONENTS

plots the forecast components.

CANDIDATES

plots model and error graphics for each candidate model fit to the series forecast series.

ALL

specifies all of the preceding PLOT= options.

For example, PLOT=FORECASTS plots the forecasts for each series.

PRINT=option | (options )

specifies the printed output desired. By default, the HPFENGINE procedure produces no printed output.

The following printing options are available:

ESTIMATES

prints the results of parameter estimation.

FORECASTS

prints the forecasts.

STATISTICS

prints the statistics of fit.

SUMMARY

prints the forecast summary.

PERFORMANCE

prints the performance statistics for each forecast.

PERFORMANCESUMMARY

prints the performance summary for each BY group.

PERFORMANCEOVERALL

prints the performance summary for all of the BY groups.

SELECT

prints the label and fit statistics for each model in the selection list.

BIAS

prints model bias information.

COMPONENTS

prints forecast model components.

CANDIDATES

prints parameter estimates for each candidate model fit to the series forecast series.

DESCSTATS

prints descriptive statistics the series forecast series.

ALL

is the same as specifying PRINT=(ESTIMATES SELECT FORECASTS STATISTICS BIAS DESCSTATS). PRINT=(ALL CANDIDATES COMPONENTS PERFORMANCE PERFORMANCESUMMARY PERFORMANCEOVERALL) prints all the options listed.

PRINTDETAILS

specifies that output requested with the PRINT= option be printed in greater detail.

REPOSITORY=catalog-name

is a two-level SAS catalog name that specifies the location of the model repository. The REPOSITORY= option can also be specified as MODELREPOSITORY=, MODELREP=, or REP=. The default for this option is SASHELP.HPFDFLT.

SCOREREPOSITORY=catalog-name | libref

is a two-level SAS catalog name that specifies the location of the model score repository. This repository is where score files are written if the SCORE statement is used in the HPFENGINE procedure. There is no default score repository. The presence of a SCORE statement requires that the SCOREREPOSITORY= option also be present.

SEASONALITY=n

specifies the length of the seasonal cycle. For example, SEASONALITY=3 means that every group of three observations forms a seasonal cycle. The SEASONALITY= option is applicable only for seasonal forecasting models. By default, the length of the seasonal cycle is 1 (no seasonality) or the length implied by the INTERVAL= option specified in the ID statement. For example, INTERVAL=MONTH implies that the length of the seasonal cycle is 12.

SORTNAMES

specifies that the variables specified in the FORECAST statement be processed in sorted order.

TASK=option

controls the model selection and parameter estimation process. Available options are as follows:

SELECT

performs model selection, estimates parameters of the selected model, and produces forecasts. This is the default.

SELECT(options)

performs model selection, estimates parameters of the selected model, produces forecasts, and potentially overrides settings in the model selection list. If a selection list does not specify a particular item and that item is specified with a TASK=SELECT option, the value as set in TASK=SELECT is used. If an option is specified in selection list, the corresponding value set in TASK=SELECT is not used unless the OVERRIDE option is also present. The available options for TASK=SELECT are as follows:

HOLDOUT=n

specifies the size of the holdout sample to be used for model selection. The holdout sample is a subset of actual time series that ends at the last nonmissing observation.

HOLDOUTPCT=number

specifies the size of the holdout sample as a percentage of the length of the time series. If HOLDOUT=5 and HOLDOUTPCT=10, the size of the holdout sample is , where is the length of the time series with beginning and ending missing values removed.

CRITERION=option

specifies the model selection criterion (statistic of fit) to be used to select from several candidate models. The following list shows the valid values for the CRITERION= option and the statistics of fit these option values specify:

SSE

sum of square error

MSE

mean square error

RMSE

root mean squared error

UMSE

unbiased mean squared error

URMSE

unbiased root mean squared error

MAXPE

maximum percent error

MINPE

minimum percent error

MPE

mean percent error

MAPE

mean absolute percent error

MDAPE

median absolute percent error

GMAPE

geometric mean absolute percent error

MINAPES

minimum absolute error percent of standard deviation

MAXAPES

maximum absolute error percent of standard deviation

MAPES

mean absolute error percent of standard deviation

MDAPES

median absolute error percent of standard deviation

GMAPES

geometric mean absolute error percent of standard deviation

MINPPE

minimum predictive percent error

MAXPPE

maximum predictive percent error

MPPE

mean predictive percent error

MAPPE

symmetric mean absolute predictive percent error

MDAPPE

median absolute predictive percent error

GMAPPE

geometric mean absolute predictive percent error

MINSPE

minimum symmetric percent error

MAXSPE

maximum symmetric percent error

MSPE

mean symmetric percent error

SMAPE

symmetric mean absolute percent error

MDASPE

median absolute symmetric percent error

GMASPE

geometric mean absolute symmetric percent error

MINRE

minimum relative error

MAXRE

maximum relative error

MRE

mean relative error

MRAE

mean relative absolute error

MDRAE

median relative absolute error

GMRAE

geometric mean relative absolute error

MAXERR

maximum error

MINERR

minimum error

ME

mean error

MAE

mean absolute error

MASE

mean absolute scaled error

RSQUARE

R-square

ADJRSQ

adjusted R-square

AADJRSQ

Amemiya’s adjusted R-square

RWRSQ

random walk R-square

AIC

Akaike information criterion

AICC

finite sample corrected AIC

SBC

Schwarz Bayesian information criterion

APC

Amemiya’s prediction criterion

INTERMITTENT=number

specifies a number greater than one that is used to determine whether or not a time series is intermittent. If the average demand interval is greater than this number, then the series is assumed to be intermittent.

SEASONTEST=option

specifies the options related to the seasonality test.

The following values for the SEASONTEST= option are allowed:

NONE

no test

(SIGLEVEL=number)

significance probability value to use in testing whether seasonality is present in the time series. The value must be between 0 and 1. A smaller value of the SIGLEVEL= option means that stronger evidence of a seasonal pattern in the data is required before PROC HPFENGINE will use seasonal models to forecast the time series.

The default is SEASONTEST=(SIGLEVEL=0.01).

ALPHA=number

specifies the significance level to use in computing the confidence limits of the forecast. The ALPHA= value must be between 0 and 1. As an example, ALPHA=0.05 produces 95% confidence intervals.

OVERRIDE

forces the use of any options listed.

NOALTLIST

disables the default action and sets the forecast to missing if no models can be fit from the initial selection list. (By default, if none of the models in a selection list can be successfully fit to a series, PROC HPFENGINE returns to the selection list SASHELP.HPFDFLT.BEST and restarts the selection process.) There will be an observation in OUTSUM=, if requested, that corresponds to the variable and BY group in question, and the _STATUS_ variable will be nonzero.

MINOBS=(TREND=n )

specifies that no trend model be fit to any series with fewer than n nonmissing values. Normally the models in a selection list are not subset by trend.

Incorporation of a trend is checked only for smoothing, UCM, and ARIMA models. For the smoothing case, only simple smoothing is a non-trend model. For UCM, the absence of a slope component qualifies it as a non-trend model. For ARIMA, there must be no differencing of the dependent variable for PROC HPFENGINE to consider it a non-trend model.

The value of n must be greater than or equal to 1. The default is n = 1.

MINOBS=(SEASON=n )

specifies that no seasonal model be fit to any series with fewer observations than n multiplied by the seasonal cycle length. The value of n must be greater than or equal to 1. The default is n = 2.

MINOBS=n

specifies that any series with fewer than n nonmissing values not be fit using the models in the selection list, but instead be forecast as the mean of the observations in the series. The value of n must be greater than or equal to 1. The default is n = 1.

FIT

estimates parameters by using the model specified in the INEST= data set, then forecasts. No model selection is performed.

FIT(options)

estimates parameters by using the model specified in the INEST= data set, then forecasts, potentially overriding the significance level in the model selection list that was used to compute forecast confidence intervals. No model selection is performed. If a selection list does not specify ALPHA and ALPHA is specified in the TASK=FIT option, the value as set in TASK=FIT will be in effect. If ALPHA is specified in selection list, the corresponding value set in TASK=FIT will not be used unless the OVERRIDE option is also present. The available options for TASK=FIT are as follows:

ALPHA=number

specifies the significance level to use in computing the confidence limits of the forecast. The ALPHA= value must be between 0 and 1.

OVERRIDE

forces the use of any options listed.

UPDATE

estimates parameters by using the model specified in the INEST= data set, then forecasts. TASK=UPDATE differs from TASK=FIT in that the parameters found in the INEST= data set are used as starting values in the estimation. No model selection is performed.

UPDATE(options)

estimates parameters by using the model specified in the INEST= data set, using the parameter estimates as starting values, then forecasts, potentially overriding the significance level in the model selection list used to compute forecast confidence intervals. No model selection is performed. If a selection list does not specify ALPHA, and ALPHA is specified in the TASK=UPDATE option, the value as set in TASK=UPDATE will be in effect. If ALPHA is specified in the selection list, the corresponding value set in TASK=UPDATE will not be used unless the OVERRIDE option is also present. The available options for TASK=UPDATE are as follows:

ALPHA=number

specifies the significance level to use in computing the confidence limits of the forecast. The ALPHA= value must be between 0 and 1.

OVERRIDE

forces the use of any options listed.

FORECAST

forecasts using model and parameters specified in the INEST= data set. No parameter estimation occurs.

FORECAST(options)

forecasts using model and parameters specified in the INEST= data set, potentially overriding the significance level in the model selection list used to compute forecast confidence intervals. No parameter estimation occurs. If a selection list does not specify ALPHA, and ALPHA is specified in the TASK=FORECAST option, the value as set in TASK=FORECAST will be used. If ALPHA is specified in the selection list, the corresponding value set in TASK=FORECAST will not be used unless the OVERRIDE option is also present. The available options for TASK=FORECAST are as follows:

ALPHA=number

specifies the significance level to use in computing the confidence limits of the forecast. The ALPHA= value must be between 0 and 1.

OVERRIDE

forces the use of any options listed.