Options Related to the Input Data Sets

AGGDATA=SAS-data-set

specifies the name of the SAS data set that contains the forecasts of the aggregated time series data. Typically, the AGGDATA= data set is generated by the OUTFOR= statement of the HPFENGINE procedure. If the AGGDATA= data set is not specified, only bottom-up reconciliation is allowed.

The AGGDATA= data set must contain a proper subset, possibly empty, of the BY variables present in the DISAGGDATA= data set. Such BY variables are called AGGBY variables. See the section BY Statement for more details about BY and AGGBY variables.

See the section AGGDATA= Data Set for more details about the AGGDATA= data set.

CONSTRAINT=SAS-data-set

specifies the name of the SAS data set that contains the constraints for the reconciled series. See the section CONSTRAINT= Data Set for more details.

DISAGGDATA | DATA=SAS-data-set

specifies the name of the SAS data set that contains the forecast of the disaggregated time series data. Typically, the DISAGGDATA= data set is generated by the OUTFOR= statement of the HPFENGINE procedure.

If the DISAGGDATA= data set is not specified, the data set last opened is used. The dimensions of the DISAGGDATA= data set are greater than the dimensions of the AGGDATA= data set.

See the section DISAGGDATA= Data Set for more details.

Options Related to the Output Data Sets

FORCECONSTRAINT

specifies whether the user-specified constraints should be forced on the PREDICT variable in the OUTFOR= data set when the problem is infeasible because the constraints are incompatible with the aggregation constraint. The default is to leave the input unmodified.

OUTFOR=SAS-data-set

specifies the name of the output SAS data set that contains the reconciled values.

OUTINFEASIBLE=SAS-data-set

specifies the name of the SAS data set that contains a summary of the nodes for which reconciliation failed because of a conflict between the constraints imposed by the user and the aggregation constraint.

OUTPROCINFO=SAS-data-set

names the output data set to contain the summary information of the processing done by PROC HPFRECONCILE . When you write a program to assess the status of the procedure’s execution, it is easier to parse a data set than it is to look at or parse the SAS log.

RECDIFF

If the RECDIFF option is specified, the OUTFOR= data sets will contain a variable named RECDIFF that is the difference between the reconciled forecasts and the original forecasts.

Options Related to Error Messages

ERRORTRACE=option

specifies how often the error and warning messages should be printed to the log.

The following values are allowed:

DATASET

Messages are printed only one time at the end of the procedure run.

AGGBY

Messages are printed for each AGGBY group.

ID

Messages are printed for each ID value.

The default is ERRORTRACE=DATASET.

Options Related to the Analysis

AGGREGATE=TOTAL | AVERAGE

specifies whether the dependent variable in the AGGDATA= data set is the total sum or the average over the BY groups of the dependent variable in the DISAGGDATA= data set. The default is AGGREGATE=TOTAL.

ALPHA=n

specifies the level of the confidence limits when CLMETHOD=GAUSSIAN. The ALPHA= value must be between 0.0001 and 0.9999. When you specify ALPHA=, the upper and lower confidence limits will have a confidence level. The default is ALPHA=0.05, which produces 95% confidence intervals.

CLMETHOD=option

specifies the method used to compute confidence limits for the reconciled forecasts.

The following methods are provided:

GAUSSIAN

The confidence intervals are computed by assuming that the forecasts are approximately Gaussian.

SHIFT

The confidence intervals are computed by re-centering the original confidence intervals around the new forecasts.

The default value is CLMETHOD=SHIFT. See the section Details: HPFRECONCILE Procedure for more information about the methods of computing confidence intervals.

DIRECTION= reconciliation-direction

specifies the reconciliation direction.

The following reconciliation values are allowed:

BU

bottom-up reconciliation

TD

top-down reconciliation

If the AGGDATA= data set is not specified, only DIRECTION=BU is allowed.

The default value is DIRECTION=BU.

See the section Details: HPFRECONCILE Procedure for more information about the reconciliation directions available in PROC HPFRECONCILE.

DISAGGREGATION=DIFFERENCE | PROPORTIONS

specifies the type of loss function for top-down reconciliation.

DISAGGREGATION=PROPORTIONS is available only when all the forecasts at a given ID value share the same sign. See the section Details: HPFRECONCILE Procedure for more information about the expressions of the loss function.

The default value is DISAGGREGATION=DIFFERENCE.

IGNOREMISSF

specifies that '.F' missing values in the PREDICT variable be treated as regular '.' missing values. If the IGNOREMISSF option is not specified, a '.F' missing value is interpreted as a failed forecast, and PROC HPFRECONCILE generates '.F' missing values for all forecasting variables in the OUTFOR= data set if that value is needed for computing the reconciled forecasts. If the IGNOREMISSF option is specified, observations that correspond to '.F' missing values are considered to belong to inactive series and therefore are not included in the reconciliation process.

LOCKZERO

specifies that zero values of the PREDICT variable in the DISAGGDATA= data be considered locked equalities. This option is available only when DIRECTION=TD. When the LOCKZERO option is active, a zero value for PREDICT in the DISAGGDATA= set implies a zero value for the corresponding observation in the OUTFOR= data set. However, if constraints are specified in the CONSTRAINT= data set for that observation, these constraints have precedence over the LOCKZERO option. Note that an unlocked equality constraint in the CONSTRAINT= data also has precedence over the LOCKZERO option. Similarly, an unlocked equality whose value is zero is not converted to a locked equality, even though the LOCKZERO option is specified.

MAXITER=k

specifies the maximum number of predictor-corrector iterations performed by the interior point algorithm. The value k is an integer between 1 and the largest four-byte, signed integer, . The default value is MAXITER=100.

PREDICTONLY

specifies that only the predicted value be reconciled.

SIGN=option

specifies the sign constraint on the reconciled series.

Valid values are as follows:

MIXED

if the output series can have any sign. This is the default.

NONNEGATIVE | POSITIVE

if the output series are supposed to be nonnegative.

NONPOSITIVE | NEGATIVE

if the output series are supposed to be nonpositive.

STDMETHOD=option

specifies the method used to compute standard errors for the reconciled forecasts.

The following methods are provided:

UNCHANGED

Reconciled standard errors are the original standard errors.

AGG

Reconciled standard errors are proportional to the original aggregated standard errors.

DISAGG

Reconciled standard errors are proportional to the original disaggregated standard errors.

The default values are STDMETHOD=UNCHANGED. See the section Details: HPFRECONCILE Procedure for more information about the methods of computing standard errors.

STDDIFBD=n

specifies a positive number that defines boundaries for the percentage difference between the original standard error and the reconciled standard error. If the percentage difference is greater than the values specified in the STDDIFBD= option, the reconciled standard error will be equal to the boundary value. For example, if STDDIFBD=0.3, the reconciled standard errors will be within a 30% band of the original standard errors.

The default value is STDDIFBD=0.25.

WEIGHTED

specifies that the loss function for top-down reconciliation be weighted by the inverse of the variance of the input forecasts.