Utility Package
UTLSTAT Object
The UTLSTAT object conveniently computes a number of forecast fit statistics for an ad hoc pair of user-specified actual and predicted time series. The computed forecast fit statistics are automatically stored in a CAS table. For each ad hoc pair of actual and predicted time series that is input into a UTLSTAT collector object, a single row of forecast fit statistics is added to the underlying CAS table. The CAS table schema that is used by the UTLSTAT object is compatible with the schema used by the HPFENGINE procedure for its OUTSTAT data set.
Table 7 shows the contents of the UTLSTAT object. For more information about the HPFENGINE procedure, see SAS Forecast Server Procedures: User's Guide.
Table 7: Contents of the UTLSTAT Object
| Column | Type | Description |
|---|---|---|
| _NAME_ | String | Variable name of actual series |
| _MODEL_ | String | Variable name of predicted series |
| DFE | Numeric | Degrees of freedom error |
| N | Numeric | Number of observations |
| NOBS | Numeric | Number of observations used |
| NMISSA | Numeric | Number of missing actuals |
| NMISSP | Numeric | Number of missing predicted values |
| NPARMS | Numeric | Number of model parameters |
| TSS | Numeric | Total sum of squares |
| SST | Numeric | Corrected total sum of squares |
| SSE | Numeric | Sum of square error |
| MSE | Numeric | Mean square error |
| RMSE | Numeric | Root mean square error |
| UMSE | Numeric | Unbiased mean square error |
| URMSE | Numeric | Unbiased root mean square error |
| MAPE | Numeric | Mean absolute percentage of error |
| MAE | Numeric | Mean absolute error |
| RSQUARE | Numeric | R-square |
| ADJRSQ | Numeric | Adjusted R-square |
| AADJRSQ | Numeric | Amemiya’s adjusted R-square |
| RWRSQ | Numeric | Random walk R-square |
| AIC | Numeric | Akaike’s information criterion |
| AICC | Numeric | Finite sample corrected Akaike’s information criterion |
| SBC | Numeric | Schwarz Bayesian information criterion |
| APC | Numeric | Amemiya’s prediction criterion |
| MAXERR | Numeric | Maximum error |
| MINERR | Numeric | Minimum error |
| MAXPE | Numeric | Maximum percentage of error |
| MINPE | Numeric | Minimum percentage of error |
| ME | Numeric | Mean error |
| MPE | Numeric | Mean percentage of error |
| MDAPE | Numeric | Median absolute percentage of error |
| GMAPE | Numeric | Geometric mean absolute percentage of error |
| MINPPE | Numeric | Minimum predicted percentage of error |
| MAXPPE | Numeric | Maximum predicted percentage of error |
| MPPE | Numeric | Mean predicted percentage of error |
| MAPPE | Numeric | Mean absolute predicted percentage of error |
| MDAPPE | Numeric | Median absolute predicted percentage of error |
| GMAPPE | Numeric | Geometric mean absolute predicted percentage of error |
| MINSPE | Numeric | Minimum symmetric percentage of error |
| MAXSPE | Numeric | Maximum symmetric percentage of error |
| MSPE | Numeric | Mean symmetric percentage of error |
| SMAPE | Numeric | Mean absolute symmetric percentage of error |
| MDASPE | Numeric | Median absolute symmetric percentage of error |
| GMASPE | Numeric | Geometric mean absolute symmetric percentage of error |
| MINRE | Numeric | Minimum relative error |
| MAXRE | Numeric | Maximum relative error |
| MRE | Numeric | Mean relative error |
| MRAE | Numeric | Mean relative absolute error |
| MDRAE | Numeric | Median relative absolute error |
| GMRAE | Numeric | Geometric mean relative absolute error |
| MASE | Numeric | Mean absolute scaled error |
| MINAPES | Numeric | Minimum absolute error percentage of standard deviation |
| MAXAPES | Numeric | Maximum absolute error percentage of standard deviation |
| MAPES | Numeric | Mean absolute error percentage of standard deviation |
| MDAPES | Numeric | Median absolute error percentage of standard deviation |
| GMAPES | Numeric | Geometric mean absolute error percentage of standard deviation |
Table 8 summarizes the methods that are associated with the UTLSTAT object.
Table 8: Methods of the UTLSTAT Object