Go to primary content
Oracle® Retail Demand Forecasting Cloud Service Implementation Guide
Release 19.0 for Windows
F24923-16
  Go To Table Of Contents
Contents

Previous
Previous
 
 

F Appendix: Configuring Batch Run

This appendix describes how to configure a batch run for RDF Cloud Service. Currently RDF CS has a Daily batch configured to export forecast. There has been inquiry about additional functionalities to be included in daily batch such as applying promotion effects during the week and producing forecast for new items that were set up during the week.

Requirements

The following requirements can be satisfied by configuring the weekly batch to run daily. However there are several issues that an implementor must pay attention before letting batch_weekly run daily.

  • The actual sales can be updated daily or weekly depends on set up. The current week's sale history will only zero or partial week. The current week's sales history should not be used in any process. When running daily batch every day, make sure your forecast start date is same as your last weekly batch forecast.

  • Pre-processing does not need to be re-run for daily and it can be turned off using the runpreprocess measure

  • The New Item batch can be optional depending on your business requirements. If there is no need to provide forecast for newly identified item during the week, it can be turned off.

  • Users may have started manually adjusting the forecast in the forecast review workbook since the last weekly batch run, it is up to the implementor to decide if they want the daily batch result to override the user's adjustment. It will be a complex issue for the implementor if the user adjustment need to be preserved.

Configuration

The following process shows how to use extensibility to configure the weekly batch to run daily.

  1. Perform the following activities for Step 1.

    1. Create a custom scalar boolean measure c_batchDailyB to distinguish Daily from the Weekly Batch Run.

    2. Configure a custom rule to calculate the c_batchDailyB. The expression can be written as

      c_batchDailyB= position([clnd].[dow] ,index([clnd].[dow], now))!="SUN"

      If today is Sunday, c_batchDailyB is false, else c_batchDailyB is true.

      When c_batchDailyB is true, daily batch should be invoked, else weekly batch is invoked.

      This rule should be store a custom rule group that is invoked in the hook_pre_load.

  2. Perform the following activity only when c_batchDailyB is true.

    1. Create custom measures and rule group to back up the following forecast parameter values:

      deffrcststartdt07 : c_deffrcststartdt07=deffrcststartdt07
               defadjfrcstmth07 : c_defadjfrcstmth07=defadjfrcstmth07
               adjfrcstmthint07 : c_adjfrcstmthint07=adjfrcstmthint07
               adjfrcstmthovr07 : c_adjfrcstmthovr07=adjfrcstmthovr07    
               runpreprocess : c_runpreprocess = runpreprocess
               runnewitembatch: c_runnewitembatch = runnewitembatch
      

      The previous expressions should be put in a rulegroup that is run in the hook_pre_preprocess

    2. Set the previous parameters to implementer desired values, such as:

      deffrcststartdt07 : deffrcststartdt07=frcststartdt07.min
               defadjfrcstmth07 : defadjfrcstmth07= 1( No adjustment) if the implementor intended for daily batch result to override all forecast adjustment
                                  defadjfrcstmth07= 2( Keep Last Change ) if the implementor intended for previous forecast adjustment to preserve.
                                  defadjfrcstmth07= 4( Keep Last Baseline ) if the implementor intended for previous baseline adjustment to preserve.
                                  defadjfrcstmth07= 5( Keep Last PEAK ) if the implementor intended for previous PEAK adjustment to preserve.                                          
               adjfrcstmthint07 : adjfrcstmthint07 = measureNA(adjfrcstmthint07)
               adjfrcstmthovr07 : adjfrcstmthovr07 = measureNA(adjfrcstmthovr07) 
           adjfrcstmthin07 and adjfrcstmthov07 modification is necessary if implementor want to keep the same adjustment principle for all timeseries.         
               runpreprocess : runpreprocess = false 
               runnewitembatch: runnewitembatch = false if new item forecast update during is week is not need. 
                            runnewitembatch = true if new item forecast update during is week is needed.
      

    The previous expressions should be put in a rulegroup that is run after Step 2a in the hook_pre_preprocess.

  3. Perform the following activities for Step 3.

    When c_batchDailyB is true, restore the forecast parameter values after the daily batch is executable

    deffrcststartdt07 : deffrcststartdt07=c_deffrcststartdt07
             defadjfrcstmth07 : defadjfrcstmth07=c_defadjfrcstmth07
             adjfrcstmthint07 : adjfrcstmthint07=c_adjfrcstmthint07
             adjfrcstmthovr07 : adjfrcstmthovr07=c_adjfrcstmthovr07    
             runpreprocess : runpreprocess = c_runpreprocess
             runnewitembatch: runnewitembatch = c_runnewitembatch
    

    The previous expressions should be put in a rulegroup that is run in the hook_pre_preprocess.