Univariate Model Forecasting was used to develop a forecast of the multiprocessor CPU busy value that was collected for system TSO1 in the system workload and performance profile case study introduced in Section 3.3.3.5. Refer to the Resource Element File Contents Report in Figure 3-92 to consider the historical observations for the CPUMPBS data element. Since the system being studied processes a random arrival testing workload, significant week to week variations around the central trend represented in the data were observed. Therefore, a geometric moving average to smooth the historical data was used so that a linear model for the observations could be developed. Figure 7-21 shows the control parameters used to produce the model.
/------------------------ Univariate Model Forecasting -----------------------\ |Command ===> | | | |Composing CA MICS Inquiry: WKLUVM - Univariate Modeling Inquiry | | | |Report title ===> Univariate Modeling Inquiry | | | |Selection criteria: | | Input file ===> WKL - System Workload Example | | Start date ===> _________ (ddmonyyyy) | | SYSID ===> TSO1 | | Zone ===> 1________ | | No. weeks ===> ____ (1-9999) | | Save forecast ===> YES (YES/NO/AGE) | | Specify CAPAPU Values ===> N (Y/N) | |Forecast data element ===> CPUMPBS | |Forecast length (weeks) ===> 8 (1-9999) | |Confidence limits (percent) ===> __ (70/90/95) | | | |Time element terms: Linear ===> YES (YES/NO) | | Quadratic ===> ___ (YES/NO) | | Cubic ===> ___ (YES/NO) | |Specify extended options ===> Y (Y/N) | \------------------------------------------------------------------------------/ Figure 7-21. System Univariate Model Forecasting (Screen 1 of 3) /--------------------------- 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 7-21. System Univariate Model Forecasting (Screen 2 of 3) /--------------------- Transform Historical Observations ---------------------\ |Command ===> | | | |Inquiry Step: Univariate Model Forecasting | | | |Smoothing technique ===> GMA (GMA/LOG10/USER/NONE) | | | | | | | |Geometric moving average data smoothing parameters: | | Equation alpha value ===> .5 (.01 - .99) | | Initial value ===> ________ | | | | | | | |User defined data transformation SAS statements: | | Statements to transform historical data ===> N (Y/N/R) | | Stmts. to reverse history transformations ===> N (Y/N/R) | | | | | | | | | | | \------------------------------------------------------------------------------/ Figure 7-21. System Univariate Model Forecasting (Screen 3 of 3)
Figure 7-22 shows the Residual Analysis Report for the model. As you can see, all of the residual values are within 10 percent of the observed values except for the weeks of July 4th and September 5th. Since these two weeks represent holidays, you can reasonably expect to encounter significant residual values for these weeks.
UNIVARIATE MODELING RESIDUAL ANALYSIS UNIVARIATE FORECAST OF: CPUMPBS BASED ON TIME ELEMENTS: LINEAR TRANSFORMED* PREDICTED RESIDUAL -------------PLOT OF RESIDUALS------------- CONFIDENCE DATE CPUMPBS CPUMPBS CPUMPBS CPUMPBS -10 0.0 +10 LIMITS +/- ------- ---------- ------------ ---------- ---------- ------------------------------------------- ---------- 24APR98 58.67 . 59.92 -1.25 | --0 | 2.44 02MAY98 64.39 61.53 60.26 4.13 | 0++++++++ | 2.30 09MAY98 63.85 64.12 60.59 3.26 | 0++++++ | 2.15 16MAY98 63.32 63.58 60.93 2.39 | 0++++ | 2.01 23MAY98 59.10 61.21 61.26 -2.17 | ----0 | 1.88 30MAY98 59.90 59.50 61.60 -1.69 | ---0 | 1.75 06JUN98 57.88 58.89 61.93 -4.06 | --------0 | 1.63 13JUN98 59.28 58.58 62.27 -2.99 | -----0 | 1.52 20JUN98 59.18 59.23 62.61 -3.43 | ------0 | 1.42 27JUN98 66.55 62.87 62.94 3.61 | 0+++++++ | 1.33 04JUL98 53.33 59.94 63.28 -9.94 | -------------------0 | 1.26 11JUL98 69.88 61.61 63.61 6.27 | 0++++++++++++ | 1.20 18JUL98 67.23 68.56 63.95 3.28 | 0++++++ | 1.17 25JUL98 71.94 69.58 64.28 7.66 | 0+++++++++++++++ | 1.16 01AUG98 64.10 68.02 64.62 -0.52 | -0 | 1.18 08AUG98 63.35 63.73 64.95 -1.60 | ---0 | 1.22 15AUG98 64.48 63.91 65.29 -0.81 | -0 | 1.28 22AUG98 63.33 63.90 65.62 -2.30 | ----0 | 1.36 29AUG98 65.86 64.59 65.96 -0.10 | 0 | 1.45 05SEP98 76.29 71.08 66.30 10.00 | 0+++++++++++++++++++ | 1.56 12SEP98 60.00 68.15 66.63 -6.63 | -------------0 | 1.68 19SEP98 66.75 63.38 66.63 0.12 | 0 | 1.68 26SEP98 67.29 67.02 66.97 0.32 | 0 | 1.80 03OCT98 69.33 68.31 67.30 2.03 | 0++++ | 1.93 10OCT98 64.37 66.85 67.64 -3.26 | ------0 | 2.07 17OCT98 70.13 67.25 67.97 2.15 | 0++++ | 2.21 24OCT98 . . 68.31 . | 0 | 2.35 31OCT98 . . 68.64 . | 0 | 2.50 07NOV98 . . 68.98 . | 0 | 2.65 14NOV98 . . 69.31 . | 0 | 2.80 21NOV98 . . 69.65 . | 0 | 2.96 28NOV98 . . 69.98 . | 0 | 3.11 05DEC98 . . 70.32 . | 0 | 3.27 12DEC98 . . 70.66 . | 0 | 3.42 * - DATA SMOOTHED USING A GEOMETRIC MOVING AVERAGE. ALPHA = 0.50
Figure 7-22. Univariate Model of System CPU Utilization
|
Copyright © 2014 CA.
All rights reserved.
|
|