Java Reference
In-Depth Information
7.4.3
Data Specification: CUSTOMERS and Their Product
Purchase Data
In this example, to explain the association concepts, we consider
five customers and three product purchase details out of possibly
millions of customers. Table 7-10 illustrates the data for these five
customer accounts, where each case is represented as multiple
records (rows). A case here corresponds to the set of products the
customer uses and a value indicating average annual balance range.
For example, Customer 1 has three records that represent three
products used and his monthly average balance range. This type of
data format is known as multirecord case or transactional format ,
where the Customer Id column is the case identifier, the attribute
name column represents the product name, and the value column
contains the average annual balance range.
7.4.4
Fine-Tune Settings: Filter Rules Based on Rule
Quality Metrics
In practice, business data can have thousands of products and
millions of product transactions. As a result, an association model
can derive a large number of rules. Rule quality thresholds are used
Table 7-10
Customer product transactions
Customer ID
Attribute Name
Amount Range
1
Checking Account
$2,000 to $5,000
1
Savings Account
$10,000 to $15,000
1
Certificate Account
$2,500 to $15,000
2
Savings Account
$10,000 to $15,000
2
Certificate Account
$2,500 to $15,000
3
Checking Account
$2,000 to $5,000
3
Savings Account
$10,000 to $15,000
4
Checking Account
$10,000 to $15,000
4
Money Market Account
$5,000 to $10,000
5
Savings Account
$10,000 to $15,000
5
Certificate Account
$2,500 to $15,000
Search WWH ::




Custom Search