Product Recommendation
In this use case, we focus on identifying what products could be recommended to a particular customer based
on similar buying trends and other historic facts.
Automated Analytics
Building the Rules
Launch SAP Predictive Analytics
From the home screen Choose the
Modeler
Section
a. Select
Create Association Rules
In
Select a
Data Source
screen enter the following values:
Data Type
: Text Files
Folder
: <input files location>
Data S
et: <Browse> select CUSTOMERS.csv
Next
to
Data Description,
select
Open Description
to open saved
Description
file.
Browse and select CUSTOMERS_DESC.txt
or
Select
Analyze
and set the following values:
CUST_NO Value to nominal
CUST_NO Key to 1, Order to 1
IF CUST_NO has not already been set to 1, a key is automatically generated. When that occurs,
set the following value:
KxIndex Key to 0
Next
to
Events Data Source
screen, enter:
Data Type
: Text Files
Folder
: <input files location>
Events
: TRANSACTIONS.csv
Next
to
Data Description, b
rowse and select TRANSACTIONS_DESC.txt
or
Select
Analyze
and set the following values:
CUST_NO Value to nominal
CUST_NO Key to 1, Order to 1
PRODUCT_NO Value to nominal
IF CUST_NO has not already been set to 1, a key is automatically generated.
When that occurs, set following value:
KxIndex Key to 0
Next
to
Association Rules Extraction Parameters
screen, enter:
Columns for Join
: Transaction
Data Set
: CUST_NO
Reference Data Set:
CUST_NO
Item Column
: PRODUCT_NO
Minimum Support
: 1
Minimum Confidence
: 20
Next
Generate
Expect a Number of Rules Found will have a value, for example 694. When your value is 0, try
lowering the values for
Minimum Support
and
Minimum Confidence
Next
to
Display
Choose
Association Rules Exploration
Select
Search
and right-click to
Add Lift
