7. UNIVARIATE MODEL FORECASTING › 7.5 Component Operation › 7.5.1 Univariate Model Forecasting Control Screen
7.5.1 Univariate Model Forecasting Control Screen
Univariate Model Forecasting control parameters allow you to
develop regression-based forecast models. You can save the
forecasts produced by the program in a forecast file that
corresponds to the resource or business element file you
specify as input to the model.
Prior to seeing the Univariate Model Forecasting screen shown
in Figure 7-8, you must follow a short set of menus which
are more fully described in Section 1.5 of this guide. The
Univariate Model Forecasting control screen is displayed
after you indicate that you want to insert a new Univariate
Model Forecast inquiry into your catalog of inquiries. Two
fields are filled in on the fourth line when you first see
this screen: inquiry name and inquiry description, separated
by a dash. You entered these two fields on the Insert
Capacity Planning Analysis screen when you requested that a
new inquiry be added.
/------------------------ Univariate Model Forecasting -----------------------\
|Command ===> |
| |
|Composing CA MICS Inquiry: TSTUVM - Univariate Modeling Inquiry |
| |
|Report title ===> Univariate Modeling Inquiry |
| |
|Selection criteria: |
| Input file ===> R02 - RESOURCE ELEM FILE WITH CAPAPU |
| Start date ===> _________ (ddmonyyyy) |
| SYSID ===> ____ |
| Hour ===> _________ |
| No. weeks ===> ____ (1-9999) |
| Save forecast ===> YES (YES/NO/AGE) |
| Specify CAPAPU Values ===> N (Y/N) |
|Forecast data element ===> PPATCBTM |
|Forecast length (weeks) ===> 10 (1-9999) |
|Confidence limits (percent) ===> 70 (70/90/95) |
| |
|Time element terms: Linear ===> YES (YES/NO) |
| Quadratic ===> YES (YES/NO) |
| Cubic ===> YES (YES/NO) |
| Create CSV Output ===> N (Y/N) |
|Specify extended options ===> N (Y/N) |
\------------------------------------------------------------------------------/
Figure 7-8. Univariate Model Forecasting Control Parameters
Some of the parameters in the Univariate Model Forecasting
screen shown in Figure 7-8 are required and some are
optional. The following is a discussion of the various
parameters:
Report The 60-character title field for the Univariate
title: Model Forecasting reports. No double quotes
(") or semicolons (;) may appear within the
body of the title.
Input file: The resource element or business element
Capacity Planning database file containing the
historical observations to be used as input to
the analysis. If you are uncertain of the
files or file names available in your Capacity
Planning database, type a question mark (?) in
the field and a selection list will be
displayed.
Start date: The start date for date range selection. If
Start date is specified, only historical
observations at least as recent as this value
are included in the analysis. The discussion
of the following control parameter describes
the options for controlling date range
selection.
SYSID: The SYSID for which the report is to be
produced. If the Capacity Planning database
file was defined as having a composite SYSID,
this value is automatically supplied on the
screen.
Hour: The Zone, Hour, or Dayname to be included in
the analysis, depending on whether you defined
the Capacity Planning database file at the
Zone, Hour, or Dayname summarization level. If
you defined the file as having a composite
Zone, Hour, or Dayname, this value is supplied
automatically on the screen. Note that the
field name will be changed automatically to
either Zone, Hour, or Dayname once you specify
the input file to reflect the input file
summarization level.
No. days/ The maximum number of days, weeks, or months of
weeks/ historical observations that will be included
months: in the analysis. Note that once you specify
the input file, the field name will change to
reflect the timespan of the Input file. When
used in conjunction with the Start date
parameter, several options are available for
date range selection control:
o If you specify only Start date, all
historical observations at least as recent
as the value specified for Start date will
be selected.
o If you specify both Start date and No.
dy/wk/mn, the date range for data selection
begins with the value specified for Start
date and extends for the specified number of
days, weeks, or months.
o If you specify only No. dy/wk/mn, the date
range for data selection will include only
this number of the most recent days, weeks,
or months.
o If you specify neither parameter, then all
historical observations will be selected.
Save Three options for saving the forecasted
Forecast: observations:
o Specifying YES causes the forecasted
observations to be saved in the Capacity
Planning database forecast file (CPFxxx01)
without aging the file, as described in
Chapter 3.
o Specifying NO results in the forecasted
observations not being saved in the forecast
file. If you select this option, the
forecast file will not be aged.
o Specifying AGE causes the forecasted
observations to be saved in the forecast
file and causes the forecast files to be
aged, as described in Chapter 3.
Specify Put a Y in this field to specify the CAPAPU
CAPAPU value(s) to be included in the analysis. The
Values CAPAPU value is specified on the CAPAPU
Selection Values screen, shown in Figure 7-9,
that appears if you enter a Y in this field.
You may specify CAPAPUs only is the Capacity
Planning database file is defined as
containing CAPAPU(s).
Forecast The data element that is to be the subject of
data Univariate Model Forecasting. If you are
element: uncertain of the correct name of the data
elements in the input file, type a question
mark (?) in this field and a selection list of
available data elements will be presented.
Forecast The length of forecast period in days, weeks,
length or months. Note that the field name changes,
day/week/ once you have specified the Input file, to
month: reflect the timespan of the Input file.
(Generally, you should not specify a forecast
length greater than the length of your
historical data series. See Section 6.2 for
guidelines on choosing this value.)\
Confidence The degree of confidence to be used in
limits calculating the upper and lower confidence
(percent): limits for the forecasted values of the
forecast data element. The default is 95%.
Time element The time elements to be used in the regression
terms: process. You must specify at least one of the
following terms: LINEAR, QUADRATIC, or CUBIC.
You may specify models of the forms shown in
Equations 7, 8, and 9 (see Section 7.4.1). For
example, you could develop a model of CPUHRS
based on linear and quadratic time terms by
specifying LINEAR and QUADRATIC. The model you
can develop using these terms corresponds to
Equation 8.
You could also develop a model of CPUHR based
on a cubic time term, corresponding to a
special case of Equation 9, where m1 and m2 are
zero.
Create CSV Default value is "N" for NO. Specify "Y" for
Output: YES to cause a Q&R formatted CSV to be created
that contains those data elements used in the
analysis. Specify "N" for NO to bypass CSV
creation.
Specify The specification for extended options.
extended Specify a Y for the opportunity to specify the
options: options on the Extended Options Menu (Figure
7-10).
If you entered a Y in the Specify CAPAPU Values field on the
Univariate Model Forecasting screen, the CAPAPU Selection
Values screen displays, as shown in Figure 7-9. One the
CAPAPU Selection Values screen, you may specify the CAPAPU
values (up to nine) to be included in the analysis. You may
specify CAPAPUs only if the Capacity Planning database file
is defined as containing the CAPAPU data elements. You may
use wild cards (? and *) when specifying CAPAPUs. A question
mark (?) replaces one character; an asterisk (*) replaces
all the characters after that point.
/-------------------------- CAPAPU Selection Values -------------------------\
|Command ===> |
| |
|Composing CA MICS Inquiry: TSTUVM Univariate Modeling Inquiry |
| |
| Rule ----+----1----+----2----+----3----+----4----+----5----+----6----|
|CAPAPU ===> ________________________________________________________________|
|CAPAPU 2 ===> ________________________________________________________________|
|CAPAPU 3 ===> ________________________________________________________________|
|CAPAPU 4 ===> ________________________________________________________________|
|CAPAPU 5 ===> ________________________________________________________________|
|CAPAPU 6 ===> ________________________________________________________________|
|CAPAPU 7 ===> ________________________________________________________________|
|CAPAPU 8 ===> ________________________________________________________________|
|CAPAPU 9 ===> ________________________________________________________________|
| |
\------------------------------------------------------------------------------/
Figure 7-9. CAPAPU Selection Values
If you entered a Y in the Specify Extended Options field on
the Univariate Model Forecasting screen, the Extended Options
Menu appears, as shown in Figure 7-10.
/--------------------------- Extended Options Menu -------------------------\
|Command ===> |
| |
|Inquiry Step: Univariate Model Forecasting |
| |
|Print Reports ===> YES (YES/NO) |
|Maximum R-Square for report ===> 0.55 (0.00-1.00) |
|P-Value flag ===> 0.05 (0.000-1.000) |
| |
|Delete historical data observations ===> N (Y/N/R) |
|Transform historical observations ===> N (Y/N/R) |
| |
\------------------------------------------------------------------------------/
Figure 7-10. Extended Options Menu
Print The Print Reports parameter controls the
Reports: printing of the Analysis of the Model and
Residual Analysis reports. If you type YES,
these reports are produced. If you type NO and
the forecasted file is NOT a Business Element
File, these reports are suppressed. These
reports will always be produced for Business
Element Files; however, the R-Square Ranking
Report is not produced.
Maximum The Maximum R-Square for report parameter is
R-Square used during the production of the R-Square
for report: Ranking Report to control the flagging of
particular forecasts. If the R-Square of a
generated model is less than this value, it
will be flagged with an RS-WARN flag. A value
of 1.0 in this field will disable the
production of the RS-WARN flag in the R-Square
Ranking Report.
P-Value The P-Value Flag is also used in the production
flag: of the R-Square Ranking Report to determine if
any of the P-values of the model variables
should be flagged with the P-WARN flag. If any
P-value in the model is greater than this
value, it will be flagged in the report. A
value of 1.0 in this field will disable the
production of the P-WARN flag in the R-Square
Ranking Report.
Note: The RS-WARN flag takes precedence over the P-WARN
flag. This means that if a model exceeds both of these
warning criteria, the RS_WARN flag will be printed.
The other options shown on the Extended Options Menu are
discussed in the following sections:
1 - Delete Selected Historical Data Observations
2 - Transform Historical Observations
Copyright © 2014 CA.
All rights reserved.
|
|