W W@ lAWlAWWA

WW wW 1








WWWW w.W


$%& &ODVVLILFDWLRQ
W W@ lAWlAWWA
WW wW 2
@

Copyright 2002 SAP AG. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or for any purpose
without the express permission of SAP AG. The information contained herein may be changed
without prior notice.

Some software products marketed by SAP AG and its distributors contain proprietary software
components of other software vendors.

Microsoft, WINDOWS, NT, EXCEL, Word, PowerPoint and SQL Server are registered
trademarks of Microsoft Corporation.

IBM, DB2, DB2 Universal Database, OS/2, Parallel Sysplex, MVS/ESA, AIX, S/390,
AS/400, OS/390, OS/400, iSeries, pSeries, xSeries, zSeries, z/OS, AFP, Intelligent Miner,
WebSphere, Netfinity, Tivoli, Informix and Informix Dynamic ServerTM are trademarks of
IBM Corporation in USA and/or other countries.

ORACLE is a registered trademark of ORACLE Corporation.

UNIX, X/Open, OSF/1, and Motif are registered trademarks of the Open Group.

Citrix, the Citrix logo, ICA, Program Neighborhood, MetaFrame, WinFrame,
VideoFrame, MultiWin and other Citrix product names referenced herein are trademarks of
Citrix Systems, Inc.

HTML, DHTML, XML, XHTML are trademarks or registered trademarks of W3C, World Wide
Web Consortium, Massachusetts Institute of Technology.

JAVA is a registered trademark of Sun Microsystems, Inc.

JAVASCRIPT is a registered trademark of Sun Microsystems, Inc., used under license for
technology invented and implemented by Netscape.

SAP, SAP Logo, R/2, RIVA, R/3, SAP ArchiveLink, SAP Business Workflow, WebFlow,
SAP EarlyWatch, BAPI, SAPPHIRE, Management Cockpit, mySAP, mySAP.com, and other
SAP products and services mentioned herein as well as their respective logos are
trademarks or registered trademarks of SAP AG in Germany and in several other countries
all over the world. MarketSet and Enterprise Buyer are jointly owned trademarks of
SAP Markets and Commerce One. All other product and service names mentioned are the
trademarks of their respective owners.
W W@ lAWlAWWA
WW wW 3

M
WM
WW M
WW@WM
WWW M
W
W

WWW WW
WWW WWWWW
WW W W
W W@ lAWlAWWA
WW wW 4
WW
W
The L CLCLL is a frequently used analytical method to classify objects (Customers,
Products or Employees) based on a particular measure (Revenue or Profit). For example, you
can classify your customers into three classes A, B and C according to the sales revenue they
generate.
ABC classification allows you to classify your data based on specified classification rules. The
data to be classified is generated by a query in the SAP BW. The classification rules refer to a
single key figure value in your data and implicitly specify which absolute or relative key figure
values map to which classes.
WW
The data to be classified using ABC classification can be taken from any system, if that system
can extract data into SAP BW. You can apply the same ABC classification to any data that has
been extracted into SAP BW.
WW@W
You require a query in SAP BW delivering suitable objects that can be classified by ABC
classification. In particular, the query must contain the key figure to which the classification rules
can be applied.
Only numerical key figure fields can be used for the classification type LLCLL FLLC L
CLCLL LL as summation of non-numerical fields is not possible.
WWW
You specify the following for the ABC classification:
The WWW for which the classification is to be performed. This entails specifying
the characteristic values to be classified (such as LLL).
The WW W that is to form the basis for classifying the characteristic values (such as
FLL CL L LCL LLL)
The WW of the characteristic that should receive the result (L L LC)
The @W for determining the data (such as FLLCLL_ LCLC L LLL)
The WW NWW for the individual ABC classes. For example, all customers
generating a profit of 0 to 20,000 belong to class C, those generating a profit between
20,001 and 80,000 to class B, and those generating more than 80,001 to class A.














W W@ lAWlAWWA
WW wW 5
W WW

1. Create a model (see Creating, Changing, and Activating a Model).
2. Create a source for training (see Creating, Changing, and Activating a Source).
3. Perform ABC Classification.
a. To do this, select the Source; use the right-hand mouse to choose L L
CLCLL in the context menu.
b. Specify whether to run the classification or as a LCLPLLL process.
c. You will find the progress indicator at the bottom of your screen while running the ABC
Classification.
d. If the ABC Classification is successful, then the system will indicate the number records
that have been classified successfully. If there are records that could not be classified
due to missing Classification Rules, then a warning is raised.
4. Display distribution as chart
a. To view the results, position the cursor on the data source, and use the right-hand mouse
to choose V LL LCL] in the context menu.
b. A graphical chart is displayed showing the distribution of classified records. The Z"CA
shows the spectrum of ranked records. The record with maximum key figure value is
displayed on the left and the record with minimum key figure value on the right. The "
CA shows the cumulated value used for the classification. Each classification is
represented by a different color.
5. Load the prediction result into the master data in SAP BW (see Loading the Prediction Result
into SAP BW).



















W W@ lAWlAWWA
WW wW 6
@W @
The ABC Classification results can be viewed in the form as an ABC Classification Chart.
Following is an example of how the typical output for the ABC Classification would look
like.


W W@ lAWlAWWA
WW wW 7
You can use the LL LLgLL button to view the classification in the list form. Following is an
example of the classification results as a detailed list.


You can use the V CL button to return to the ABC Classification Chart.

W W@ lAWlAWWA
WW wW 8
W WW

The input data for SAPs ABC Classification is basically divided into two parts:

Model Fields
Model Parameters



WWW WW
Model fields are those fields that are part of the ABC classification. You must specify the following
three model field values.
CL L[LL. You must specify which characteristic values to be classified (for
example, Info Object LFL for Business Partners to be classified)
A CLCLL LL. You must specify the key figure field, which is used as the basis
for the classification (for example, Info Object LLL for the Sales Revenue
generated by the customer)
w C !LL[LL. You must specify the Info Object used as attribute for the classification
result (for example, Info Object LL<< representing the possible classification
values A, B, and C).

A
w
W W@ lAWlAWWA
WW wW 9
WWW WWWW
You use the WW @WWWW to specify the type of and their mappings. The
classification rules consist of a threshold and the corresponding class as shown in the table
below.
WW W
Thres_1 C_1
Thres_2 C_2
Thres_3 C_3


A threshold of in the table specifies that the left-open value interval (" ]
maps to class . denotes infinite values.
Consequently, the first row in the table specifies that all records with values less than or equal to
will be mapped to class . The second row specifies that all values less than or equal
to Z but higher than will be mapped to Z, and so on.
The referenced value depends on the specified type of classification rules.


W W@ lAWlAWWA
WW wW 10
WW W
ABC classification provides the following four types of classification rules:
o $EVROXWH 9DOXHV RI &ODVVLILFDWLRQ &ULWHULRQ+ Each record in the dataset is classified
based on the absolute value of its referred key figure value.

You need to classify your LLL according to the <C 1L they
generated last year. You require the customers with sales revenue of:

At least $ 80,000 and more to be classified as
Less than $ 80,000 but more than $ 20,000 to be classified as , and
Less than $ 20,000 to be classified as
This classification can be done using classification rule type LLLL CL L
CLCLL LL containing the corresponding mappings. Each mapping
specifies which key figure value interval (<C 1L in this case) maps to
which classification value.
o &XPXODWHG 3HUFHQWDJH RI &ODVVLILFDWLRQ &ULWHULRQ+ All data records are ranked by their
key figure value and starting with the record with maximum key figure value. In the order
of descending key figure values, each record is classified based on the ratio:
LLCLL _ L CL / LLC L _ L CL.
This is in terms of percentage. The LLCLL _ L CL denotes the sum of all
key figure values up to and including the record to be classified. The sum of key figure
values denotes the sum of all key figure values.

You want the LL gLLCL customers with highest sales revenue and together
generating sales revenue of at most 10 percent of the total sales revenue to be
classified as .
The LL gLL customers with next highest sales revenue and together
generating the next 50 percent of the total sales revenue last year are to be
classified as .
The remaining LV gLL customers with lowest sales revenue are to be classified
as .
You can do this classification by using classification rule type LLCLL
FLLC L CLCLL LL.

A single record hitting the borderline, that is, a record falling in between classes
is always assigned to the class with the higher threshold. The following example
illustrates this.
Following is the table of classification rules for an ABC classification of type
LLCLL FLLC L CLCLL LL.




W W@ lAWlAWWA
WW wW 11
WW W
10 A
40 B
100 C
The first line specifies that those customers with highest revenue and together
leading to 10 percent of total revenue map to class . If we assume that the
following three customers have to be classified according to these rules:
W WNWW
Miller 10.000 $
Smith 5.000 $
Green 5.000 $
In this case, the result of this classification is that all the three customers are
classified to class C. This is because Miller contributes 50 percent to total
revenue, Miller and Smith together contribute 75 percent of total revenue and all
three customers together contribute to 100 percent of total revenue. All
percentages of cumulated revenues are higher than 10 percent and even higher
than 40 percent and so they are all classified to class .
o &XPXODWHG 3HUFHQWDJH RI &ODVVLILHG 2EMHFW All data records are ranked by their key
figure value and starting with the top-ranked record with maximum key figure value. In the
order of descending key figure values, each record is classified based on the ratio
CP FLLL / LL L LLL
This is in terms of percentage. The CP FLLL denotes position in the key-figure-
based ranking while the top-ranked has ranking position 1; the secondly best-ranked
record has ranking position 2 and so on. The LL L LLL denotes the total
number of records to be classified.

As in the example used in type LLCLL FLLC L CLCLL LL,
you can classify your customers into top 10% as class next 50% as and the
remaining as profitability customers.
o $EVROXWH 9DOXHV RI &ODVVLILHG 2EMHFW+ All data records are ranked by their key figure
value, and starting with the top-ranked record with maximum key figure value in the
descending order. Each record is classified based on the absolute ranking position.

In this case, the top 10 records, depicting the top 10 most profitable customers
with highest sales revenue will be classified as . The next 2000 records
depicting the medium profit customers will be under class and the remaining
low profit customers generating the least sales revenue would be classified as .