Skip to Content

Tutorial of Association Analysis

Hello everyone,

I am here going to demonstrate how to use Data Mining Model of SAP BW to operate Assocation Analysis.

In order to complete this tutorial, data source CSV file which contains relevant data is needed for data extraction. Also you can create your own like this:

24-05-2013 Friday 10-26-39 PM.jpg

   Access the Data Mining Workbench.  Use TCODE RSDMWB or Double Click Data Mining Model

23-05-2013 Thursday 4-03-22 PM.jpg

  The Data Mining Workbench screen appears as follows:

23-05-2013 Thursday 4-07-08 PM.jpg

The workbench includes a list of Data Mining methods that can be used to analyze data, commonly stored in business applications.These methods include classification (decision trees), clustering, association analysis, approximation, and further analyses techniques such as ABC classification.


Phase 1  — Creating an Association Model

1. Expand 23-05-2013 Thursday 4-34-11 PM.jpgand Right Click 23-05-2013 Thursday 4-44-50 PM.jpgCreate Model,

The Create Model dialog screen appears as follows:

23-05-2013 Thursday 5-00-09 PM.jpg

2. Name Model and Description, select Radio button: Manul, and accept the entry.

The Create Model XXX screen will appear as follows:

23-05-2013 Thursday 4-57-36 PM.jpg

You now need to select the fields (InfoObjects) needed for your model. The fields selected will need to match those identified in the data source file shoppingdataset.csv.

3. Click    the icon 23-05-2013 Thursday 5-10-43 PM.jpg to insert a row for the first InfoObject (Material).

Notice that properties (data type and length) of the InfoObject are automatically populated. The Content Type option defines how each field is to be treated within the model. There are three possible options; transaction, item and transaction weight.

4. Click  23-05-2013 Thursday 5-35-28 PM.jpg  the icon  in each specific field to display all the available options and select corresponding option.

After completing this step the field assignment for the Association Model show appear as follows;

23-05-2013 Thursday 5-53-42 PM.jpg

5. Click  Parameters  tab to set the following parameters.

Support can be used to help determine how useful a product association rule is.

Confidence (or predictability) measures how dependent a particular item.

Lift helps to identify and eliminate rules that are only generated because some of the items naturally occur very frequently, although there is no actual association between the set of leading and depending items.

The Association Model uses Leading and Dependent items. If a leading item has been purchased, what is the likelihood the following dependent items is also purchased.  If a customer purchases milk (leading item) what is the likelihood they will also purchase bread (dependent item). The relationship between leading and dependent items is not limited to one to one. The parameter Leading Depth can be set to identify the number of products to be considered together as leading items and the parameter Dependent Depth can be used to set the number of dependent items to be considered in the model’s analysis.

                   Leading Items                        Dependent Items

                   milk                                           bread

                   milk, spread                            bread

                   milk                                           bread,spread

6. Put value in the Maximum Leading Depth field, Tand in the Maximum Dependent Depth field.

7. Click 16-05-2013 Thursday 2-25-22 PM.jpgreturn to the previous screen.

8. Click 16-05-2013 Thursday 2-01-11 PM.jpg to save your model.

A message appears in the Status Bar.

9. Click 23-05-2013 Thursday 6-35-20 PM.jpg to activate the model.  A log screen will appear indicating the status of each step of the model building proces23-05-2013 Thursday 6-36-57 PM.jpg

10. Click  23-05-2013 Thursday 6-36-46 PM.jpg to close this screen. Now you have successfully created Association Model

Next, I am going to use SAP GUI and run Phase 2 & 3 — Training and Execution of the Assocation Model :

Thanks, I hope you enjoy it !

You must be Logged on to comment or reply to a post.