Predictive Analytics Forecasting

How-To Guide

1 - Maintain the number of days of data to be read from the System Monitoring data store, whether to read from the System Analysis aggregated store (if aggregation is enabled) and the number of forecast points for different granularities available in the System Monitoring - Metric Monitoring and System Analysis applications (15MIN, 1HOUR, 5MIN, DAY, MIN) using SM30 view of PAS_SM_DP_CONFIG.

Note: Default is maintained automatically based on granularity during preparation.

MIN

  • Number of days of data to be read from System Monitoring data store: 5
  • Read from Aggregation Store: No
  • Number of forecast points to be shown: 12

5MIN

  • Number of days of data to be read from System Monitoring data store: 15
  • Read from Aggregation Store: No
  • Number of forecast points to be shown: 6

15MIN

  • Number of days of data to be read from System Monitoring data store: 15
  • Read from Aggregation Store: No
  • Number of forecast points to be shown: 4

1HOUR

  • Number of days of data to be read from System Monitoring data store: 30
  • Read from Aggregation Store: Yes
  • Number of forecast points to be shown: 2

DAY

  • Number of days of data to be read from System Monitoring data store: 30
  • Read from Aggregation Store: Yes
  • Number of forecast points to be shown: 2
     

2 - Execute the PAS_SAVE_EXCEPTION report through transaction SE38. By default, SAP APL is used as the algorithm. The user can adjust the algorithm to FS (SAP PAL's Forecast Smoothing) for metrics by saving the metric id and hash metric path (if applicable).

Note: Default is SAP APL.
 

3 - Maintain FORECAST_CONFIDENCE using SM30 view of PAS_SM_GEN_CONFI : Confidence Interval for SAP APL Auto Regression algorithm. The other acceptable values are 99 and 90. 

  • Param Name: FORECAST_CONFIDENCE
  • Param Count: 1
  • Param Value: 95

Note: Applicable only for SAP APL from SAP Focused Run 1.0 FP03 onwards. Default is 95%.
 

4 - Maintain NULL_VALUE_EXIT_PC using SM30 view of PAS_SM_GEN_CONFI : The percentage of null values acceptable for forecasting. If there are more null values, forecasting will be aborted.

  • Param Name: NULL_VALUE_EXIT_PC
  • Param Count: 1
  • Param Value: 40

Note: Default is 40%.