Discover how you can use time distribution in update rules to combine data with differing levels of granularity. Step through the configuration for time distribution and familiarize yourself with the concept via examples.
Key Concept
Time distribution allows you to change granularity of data and allocates data equally across time periods. This enables you to report on monthly data as if it were daily data.
You cannot combine two different data sources with different time characteristics. For example, say you load daily sales data from R/3 with a time characteristic of 0CALDAY into a sales analysis InfoCube. Management needs you to combine this sales data with data from another non-SAP R/3 source that provides only monthly sales data.
This creates a challenge when viewing data at the daily level because you must disaggregate the monthly data to daily values to create a daily analysis of both types of data. It is difficult to manually disaggregate data. You could attempt to create a custom start routine in the update rules to allocate the monthly data across days, but there is a much easier and more efficient way.
SAP BW contains an easy-to-use function to disaggregate data based on date — time distribution in update rules. This function allows SAP BW to equally distribute key figure data from a less granular source to a more granular source by taking an equal percentage of each key figure value and allocating this percentage across time. You cannot weight the values so the first part of the week receives a higher percentage than the last part of the week; the distribution is always equal across days.
The system performs time distribution for all key figure values in update rules mapped from the source data. Thus, if three key figure values are mapped, the system allocates all three key figure values across time periods. You cannot use the time distribution on a subset of key figures.
The time distribution feature is available starting with SAP BW Release 3.0B, and you can use it to allocate both SAP and non-SAP source data. In other words, it works on any data that you load into an InfoCube via update rules regardless of its source.
You can only use time distribution with standard time InfoObjects (e.g., 0CALDAY, 0CALWEEK) and you cannot use it for any custom time characteristics or with transfer rules. It works with all key figures types such as quantities and currency. You can only use this feature in update rules when loading data into an InfoCube; it cannot distribute data into an ODS.
Monthly to Weekly Time Distribution
To allocate data from a monthly source data to a weekly InfoCube, SAP BW first breaks down the key figure value into days by calculating 1/x (x = number of days in the month). Then the system examines each week in the month and determines how many of those days fall into each month. SAP BW's week starts on Monday and ends on Sunday.
Let's see how SAP BW would load actual source data from November 2005 using time distribution in update rules. In November 2005, the first day of the month starts on a Tuesday. This is week 44 of the year. The system places an equivalent value of 1/30th of the source total value (300, shown in Table 1) for each of that week's six days into week 44/2005. Then the system automatically allocates this 1/30th value for each day in the month and places any remainder in the last day of the month. The total for week 44/2005 is 1/30 x 300 (total value) x 6 (days in week) = 60, shown in Table 2. Note the smaller remainder value for week 48/2005.
|
Customer
|
Month/year
|
Value
|
|
ABC
|
11/2005
|
300
|
| Table 1 |
Total value for November 2005 |
|
Customer
|
Month/year
|
Value
|
|
ABC
|
44/2005
|
60
|
|
ABC
|
45/2005
|
70
|
|
ABC
|
46/2005
|
70
|
|
ABC
|
47/2005
|
70
|
|
ABC
|
48/2005
|
30
|
| Table 2 |
Weekly values for November 2005 |
Set Up Time Distribution
To set up time distribution, go into the update rules via Administration Workbench (transaction RSA1). Choose the InfoProvider tab, find your InfoCube, select the update rules for the InfoCube, double-click on one of the key figure values, and choose the Time Ref. tab. The system displays all time characteristics for your update rules. Click on the method icon. The last choice on the list is Time Distribution (Figure 1).
Figure 1
Set up time distribution in the Time Ref. tab of the update rules
Choose the Time Distribution button and enter a source field date value to tell the system which date characteristic to distribute from the source system. In this case, choose the Calendar Year/Mon (0CALMONTH) field to convert from calendar year/month to 0CALWEEK.
The system prompts you to choose between Normal Calendar or Factory Calendar. The normal calendar refers to the standard Gregorian calendar. Factory calendar allows the system to apply time distribution based on a custom SAP factory calendar. You can define calendars using transaction SCAL. If you use a custom calendar, the system uses the logic described in the "Monthly to Weekly Time Distribution" section to distribute the values. However, you can configure the system to use a certain number of days for each month by creating a custom calendar.
After you've selected the choices described above, the system immediately warns that time distribution is active for all key figure values (Figure 2). This means that the system will now distribute all key figure values in the update rules.
Figure 2
System warns that it will apply time distribution to all key figure values in the update rules
In the November 2005 example, let's say that you load the flat file data shown in Table 3 into the InfoCube, which has a time value of 0CALWEEK. The system equally divides the values by week and places any remainder in the final week. After you load the data into the InfoCube, you can view it using transaction LISTCUBE (Figure 3).
|
Customer
|
Month/year
|
Total sales
|
Total sales quantity |
|
ABC
|
11/2005
|
300
|
350
|
| Table 3 |
Flat file data for November 2005 |
Figure 3
Weekly data for November 2005 after time distribution
Figure 3 shows two key figures: 0SALES represents the dollar value of sales and ZSLSQTY is the total sales quantity. The system distributes the values of these two key figures based on the source month 11/2005. The 0SALES value divides evenly and the system allocates ZSLSQTY's remainder to November's last week, 200548.
Let's see how SAP BW would distribute the same data by day. To change the InfoCube's data from weekly to daily, purge the existing data and change the 0CALWEEK time distribution characteristic to 0CALDAY using the screens shown in Figure 1. If you load the same source data into this new InfoCube, your results should look like Figure 4. Note the smaller remainder value, 11.657, on 11/30/2005, the last day of the month.
Figure 4
Daily data shown after time distribution. Notice the remainder in the final day of the month.
