Previous Topic: 7.3.1 Model Analysis ReportNext Topic: 7.3.3 R-Square Ranking Report


7.3.2 Residual Analysis Report


The Residual Analysis Report shows how the model proposed by
the analyst fits the historical data series.  Figure 7-2
shows an example of a Residual Analysis Report for the
channel SIO count.

CA MICS Capacity Planner UNIVARIATE MODELING RESIDUAL ANALYSIS UNIVARIATE FORECAST OF: CHAN5CNT BASED ON TIME ELEMENTS: LINEAR TRANSFORMED* PREDICTED RESIDUAL -------------PLOT OF RESIDUALS------------- CONFIDENCE DATE CHAN5CNT CHAN5CNT CHAN5CNT CHAN5CNT -195E4 0.0 +195E4 LIMITS +/- ------- ---------- ------------ ---------- ---------- ------------------------------------------- ---------- 30SEP97 11882418 . 10324734 1557683.74 | 0+++++++++++++++ | 1624240.54 31OCT97 12860633 12371525.50 10912672 1947961.05 | 0++++++++++++++++++++| 1416789.90 28NOV97 11743581 12302107.00 11500610 242971.35 | 0++ | 1221119.32 31DEC97 12510012 12126796.50 12088547 421464.66 | 0++++ | 1043874.36 30JAN98 12071368 12290690.00 12676485 -605117.03 | ------0 | 896056.61 27FEB98 12076524 12073946.00 13264423 -1187899 | ------------0 | 794269.29 31MAR98 12186976 12131750.00 13852360 -1665384 | -----------------0 | 757306.06 30APR98 14355185 13271080.50 14440298 -85113.10 | 0 | 794269.29 29MAY98 15011392 14683288.50 15028236 -16843.79 | 0 | 896056.61 30JUN98 17176448 16093920.00 15616173 1560274.52 | 0++++++++++++++++ | 1043874.36 31JUL98 17421158 17298803.00 16204111 1217046.83 | 0++++++++++++ | 1221119.32 31AUG98 18042957 17732057.50 16792049 1250908.14 | 0++++++++++++ | 1416789.90 31SEP98 . . 17379987 . | 0 | 1624240.54 31OCT98 . . 17967924 . | 0 | 1839490.02 31NOV98 . . 18555862 . | 0 | 2060095.20 31DEC98 . . 19143800 . | 0 | 2284505.07 31JAN99 . . 19731737 . | 0 | 2511700.04 31FEB99 . . 20319675 . | 0 | 2740987.65 31MAR99 . . 20907613 . | 0 | 2971883.57 31APR99 . . 21495550 . | 0 | 3204040.12 * - DATA SMOOTHED USING A GEOMETRIC MOVING AVERAGE. ALPHA = 0.50


 Figure 7-2.  Residual Analysis Report

The Residual Analysis Report shown in Figure 7-2 presents the
following information:

UNIVARIATE
FORECAST OF: The name of the data element to be forecast.

BASED ON
TIME         The time elements (LINEAR, QUADRATIC, and/or
ELEMENTS:    CUBIC) that you are using to develop the model.

The columns of the report are discussed below:

DATE:        The ending date of the week or month that is
             represented by the observation.

CHAN5CNT:    The actual historical observations for the data
             element being modeled (in this case, CHAN5CNT).
             If the last two characters of the the data
             element name are TM, the facility assumes that
             the value is in seconds and uses a SAS TIME10.1
             format for the report.

TRANSFORMED  The transformed value.  If you do not specify
(CHAN5CNT):  that historical observations are to be
             transformed by one of the options available on
             the Transform Historical Observations screen
             shown in Figure 7-12, the TRANSFORMED value will
             be identical to the historical value.  If you do
             specify that historical observations are to be
             transformed, then a footnote is generated to
             indicate the type of processing that is
             performed on the historical data series.  The
             presence of a footnote is indicated by an
             asterisk (*) following the word TRANSFORMED.  If
             the last two characters of the the data element
             name are TM, the facility assumes that the value
             is in seconds and uses a SAS TIME10.1 format for
             the report.

PREDICTED    The y value predicted for the interval using the
(CHAN5CNT):  regression equation fit to the historical data.
             If the last two characters of the data element
             name are TM, the facility assumes that the value
             is in seconds and uses a SAS TIME10.1 format for
             the report.

RESIDUAL     The error term, e, in the regression equation.
(CHAN5CNT):  The residual is equal to the difference between
             the actual historical observation and the
             predicted value.  The residual value for a point
             deleted from the modeling process is of no
             concern, since the model was not developed to
             account for the observation.  If the last two
             characters of the data element name are TM, the
             facility assumes that the value is in seconds
             and uses a SAS TIME10.1 format for the report.

PLOT OF      The residual values that are plotted over the
RESIDUALS:   range from minus to plus the maximum residual
             value.

CONFIDENCE   The 95% confidence limits for the forecast.  As
LIMITS:      discussed in Section 7.2, the confidence limits
             bound the potential error that may exist in the
             forecasts produced by the model.  If the last
             two characters of the the data element name are
             TM, the facility assumes that the value is in
             seconds and uses a SAS TIME10.1 format for the
             report.