Previous Page | Next Page

The X12 Procedure

Example 34.7 MDLINFOIN= and MDLINFOOUT= Data Sets

This example illustrates the use of MDLINFOIN= and MDLINFOOUT= data sets. Using the data set shown, PROC X12 step identifies the model with outliers as displayed in Output 34.7.1. Output 34.7.2 shows the data set that represents the chosen model.

data b1;
   input y @@; 
   datalines;
   112 118 132 129
   121 135 148 148
   136 119 104 118
   115 126 141 135
   125 149 270 170
   158 133 114 140
;
title 'Model Identification Output to MDLINFOOUT= Data Set';
proc x12 data=b1 start='1980q1' interval=qtr MdlInfoOut=mdl;
   automdl;
   outlier;
run ;
proc print data=mdl;
run;

Output 34.7.1 Displayed Model Identification with Outliers
Model Identification Output to MDLINFOOUT= Data Set

The X12 Procedure

Critical Values to use in Outlier
Detection
For Variable y
Begin 1980Q1
End 1985Q4
Observations 24
Method Add One
AO Critical Value 3.419415
LS Critical Value 3.419415

Final Automatic Model Selection
For Variable y
Source of Model Estimated Model
Automatic Model Choice ( 2, 1, 0) ( 0, 0, 0)

Regression Model Parameter Estimates
For Variable y
Type Parameter NoEst Estimate Standard Error t Value Pr > |t|
Automatically Identified AO 1984Q3 Est 102.36589 5.96584 17.16 <.0001

Exact ARMA Maximum Likelihood Estimation
For Variable y
Parameter Lag Estimate Standard Error t Value Pr > |t|
Nonseasonal AR 1 0.40892 0.20213 2.02 0.0554
  2 -0.53710 0.20975 -2.56 0.0178

Output 34.7.2 PROC X12 MDLINFOOUT= Data Set Model with Outlier Detection
Model Identification Output to MDLINFOOUT= Data Set

Obs _NAME_ _MODELTYPE_ _MODELPART_ _COMPONENT_ _PARMTYPE_ _DSVAR_ _VALUE_ _FACTOR_ _LAG_ _SHIFT_ _NOEST_ _EST_ _STDERR_ _TVALUE_ _PVALUE_ _STATUS_ _SCORE_ _LABEL_
1 y REG EVENT SCALE AO AO01JUL1984D . . . . 0 102.366 5.96584 17.1587 0.000000 .    
2 y ARIMA FORECAST NONSEASONAL DIF y . . 1 . . . . . . .    
3 y ARIMA FORECAST NONSEASONAL AR y . 1 1 . 0 0.409 0.20213 2.0231 0.055385 .    
4 y ARIMA FORECAST NONSEASONAL AR y . 1 2 . 0 -0.537 0.20975 -2.5606 0.017830 .    


Suppose that after examining the output from the preceding example, you decide that an Easter regressor should be added to the model. The following statements create a data set with the model identified above and adds a U.S. Census Bureau Predefined Easter(25) regressor. The new model data set to be used as input in the MDLINFOIN= option is displayed in the data set shown in Output 34.7.3.

data pluseaster;
   _NAME_ = 'y';
   _MODELTYPE_ = 'REG';
   _MODELPART_ = 'PREDEFINED';
   _COMPONENT_ = 'SCALE';
   _PARMTYPE_ = 'EASTER';
   _DSVAR_ = 'EASTER';
   _VALUE_ = 25;
run;

data mdlpluseaster;
   set mdl;
run;

proc append base=mdlpluseaster data=pluseaster force;
run;
proc print data=mdlpluseaster;
run;

Output 34.7.3 MDLINFOIN= Data Set Model with Easter(25) Regression Added
Model Identification Output to MDLINFOOUT= Data Set

Obs _NAME_ _MODELTYPE_ _MODELPART_ _COMPONENT_ _PARMTYPE_ _DSVAR_ _VALUE_ _FACTOR_ _LAG_ _SHIFT_ _NOEST_ _EST_ _STDERR_ _TVALUE_ _PVALUE_ _STATUS_ _SCORE_ _LABEL_
1 y REG EVENT SCALE AO AO01JUL1984D . . . . 0 102.366 5.96584 17.1587 0.000000 .    
2 y ARIMA FORECAST NONSEASONAL DIF y . . 1 . . . . . . .    
3 y ARIMA FORECAST NONSEASONAL AR y . 1 1 . 0 0.409 0.20213 2.0231 0.055385 .    
4 y ARIMA FORECAST NONSEASONAL AR y . 1 2 . 0 -0.537 0.20975 -2.5606 0.017830 .    
5 y REG PREDEFINED SCALE EASTER EASTER 25 . . . . . . . . .    


The following statements estimate the regression and ARIMA parameters by using the model described in the new data set mdlpluseaster. The results of estimating the new model are shown in Output 34.7.4.

proc x12 data=b1 start='1980q1' interval=qtr
   MdlInfoIn=mdlpluseaster MdlInfoOut=mdl2;
   estimate;
run;

Output 34.7.4 Estimate Model with Added Easter(25) Regression
Model Identification Output to MDLINFOOUT= Data Set

The X12 Procedure

Regression Model Parameter Estimates
For Variable y
Type Parameter NoEst Estimate Standard Error t Value Pr > |t|
Easter Easter[25] Est 6.15738 4.89162 1.26 0.2219
User Defined AO01JUL1984D Est 105.29433 6.15636 17.10 <.0001

Exact ARMA Maximum Likelihood Estimation
For Variable y
Parameter Lag Estimate Standard Error t Value Pr > |t|
Nonseasonal AR 1 0.44376 0.20739 2.14 0.0443
  2 -0.54050 0.21656 -2.50 0.0210


The new model estimation results are displayed in the data set mdl2 shown in Output 34.7.5.

proc print data=mdl2;
run;

Output 34.7.5 MDLINFOOUT= Data Set, Estimation of Model with Easter(25) Regression Added
Model Identification Output to MDLINFOOUT= Data Set

Obs _NAME_ _MODELTYPE_ _MODELPART_ _COMPONENT_ _PARMTYPE_ _DSVAR_ _VALUE_ _FACTOR_ _LAG_ _SHIFT_ _NOEST_ _EST_ _STDERR_ _TVALUE_ _PVALUE_ _STATUS_ _SCORE_ _LABEL_
1 y REG PREDEFINED SCALE EASTER EASTER 25 . . . 0 6.157 4.89162 1.2588 0.22193 .    
2 y REG EVENT SCALE AO AO01JUL1984D . . . . 0 105.294 6.15636 17.1033 0.00000 .    
3 y ARIMA FORECAST NONSEASONAL DIF y . . 1 . . . . . . .    
4 y ARIMA FORECAST NONSEASONAL AR y . 1 1 . 0 0.444 0.20739 2.1397 0.04428 .    
5 y ARIMA FORECAST NONSEASONAL AR y . 1 2 . 0 -0.541 0.21656 -2.4959 0.02096 .    


Previous Page | Next Page | Top of Page