7. UNIVARIATE MODEL FORECASTING › 7.5 Component Operation › 7.5.4 SAS Statements Screens

7.5.4 SAS Statements Screens

You can use the first SAS Statements screen, shown in Figure
7-13, for coding SAS statements to transform historical data
prior to the regression analysis performed by Univariate
Model Forecasting.

You can use the second screen, shown in Figure 7-14, to
reverse this transformation process after the regression
is done, so that the reports show the untransformed values.

/--------------------------  SAS System Statements  ----------------------------\
|Command ===>                                                  Scroll ===> PAGE |
|                                                                               |
|SAS Statements for the ASGN_TRN Macro                                          |
|-------------------------------------------------------------------------------|
|****** ***************************** TOP OF DATA ******************************|
|==MSG> ENTER SAS STATEMENTS TO PERFORM A USER TRANSFORMATION.                  |
|=NOTE= %MACRO ASGN_TRN; <=== PLEASE DO NOT PLACE LINES BEFORE THIS LINE.       |
|000001                                                                         |
|000002                                                                         |
|000003                                                                         |
|000004                                                                         |
|000005                                                                         |
|000006                                                                         |
|000007                                                                         |
|000008                                                                         |
|000009                                                                         |
|000010                                                                         |
|=NOTE= %MEND ASGN_TRN; <=== PLEASE DO NOT PLACE LINES AFTER THIS LINE.         |
|****** **************************** BOTTOM OF DATA ****************************|
|                                                                               |
\------------------------------------------------------------------------------/

 Figure 7-13.  SAS Statements Screen - ASGN_TRN Macro

/--------------------------  SAS System Statements  ----------------------------\
|Command ===>                                                  Scroll ===> PAGE |
|                                                                               |
|SAS Statements for the RET_TRN Macro                                           |
|-------------------------------------------------------------------------------|
|****** ***************************** TOP OF DATA ******************************|
|==MSG> SPECIFY SAS STATEMENTS TO REVERSE A USER TRANSFORMATION.                |
|=NOTE= %MACRO RET_TRN; <=== PLEASE DO NOT PLACE LINES BEFORE THIS LINE.        |
|000001                                                                         |
|000002                                                                         |
|000003                                                                         |
|000004                                                                         |
|000005                                                                         |
|000006                                                                         |
|000007                                                                         |
|000008                                                                         |
|000009                                                                         |
|000010                                                                         |
|=NOTE= %MEND RET_TRN; <=== PLEASE DO NOT PLACE LINES AFTER THIS LINE.          |
|****** **************************** BOTTOM OF DATA ****************************|
|                                                                               |
\------------------------------------------------------------------------------/

 Figure 7-14.  SAS Statements Screen - RET_TRN Macro

The remainder of this section describes considerations for
coding these macros.

Note that the %MACRO statements appearing on each of these
screens are already provided for you and are field-protected
from modification.

Univariate Model Forecasting does not work directly with the
variable names on the input SAS file, but rather uses SAS
macro variables to pass the names taken from the control
parameters into the program's code.  This allows Univariate
Model Forecasting to be sufficiently generalized to handle
all incoming SAS files without imposing naming conventions on
you.  If you are writing SAS statements to transform the
original historical data using the first screen shown in
Figure 7-13, refer to the following symbolic names for
variables:

    &VNAME:  The original dependent variable

    _TRAN_V_:  The transformed dependent variable

In some instances, you may want to code a reverse
transformation macro that returns the predicted variables to
the original dependent variable units.  When the reverse
transformation macro is invoked, the SAS variable names that
contain the critical data are as follows:

     ___ESTS:   The predicted value of the transformed
                dependent variable (three leading
                underscores)

     _____UCL:  The upper confidence limit of the transformed
                predicted variable (five leading underscores)

     _____LCL:  The lower confidence limit of the transformed
                predicted variable (five leading underscores)

To illustrate this concept, assume there is a resource
element file that contains CPU utilization data.  Further,
assume that on March 8, 1998, a new CPU was installed that is
approximately 30% more powerful than its predecessor.  A CPU
utilization trend over time would probably show a sudden
decline in CPU utilization on or around March 8, 1998.  This
discontinuity in the trend would result in a very poor fit by
a linear, quadratic, or cubic equation.

To provide a set of historical points that can result in a
better fit to a standard equation, you can transform all the
historical data after March 8, 1998, by multiplying it by 1.3
(representing the 30% increase provided by the new
processor).  Following the curve fitting, the projected trend
is transformed back into units of the current processor by
dividing the forecast values by 1.3.  (This adjustment for
relative processor speeds can be done for you automatically
if you use the CPU Planning Standard Application described in
Chapter 3.  The process described above is for illustrative
purposes only.)

On the first screen, shown in Figure 7-13, you would enter
the following SAS statements:

    IF DATE GE '08MAR98'D THEN _TRAN_V_ = 1.3 * &VNAME;
    ELSE _TRAN_V_ = &VNAME;

On the second screen, you would enter the following SAS
statements:

    IF DATE GE '08MAR98'D THEN DO;
      ___ESTS = ___ESTS / 1.3;
      _____LCL = _____LCL / 1.3;
      _____UCL = _____UCL / 1.3;
    END;