The Mining Data Mart - Data Mining Techniques in CRM: Inside Customer Segmentation - page 144

Table 4.5 The transactions table.

Field name Field type Field description

Year_Month Int The number that indicates the year and

month of data

Customer_ID Int The customer identification number

Account_ID Int The product account identification num-

ber

Transaction_Code Int The transaction code (see next ''L_

Transaction_Code'' lookup table)

Transaction_Channel Int The transaction channel (see next ''L_

Transaction_Channel'' lookup table)

Num_Transactions Int Total number of transactions within month

Amount_Transactions Real Total amount of transactions within month

Min_Amount Real Minimum amount of transactions withinmonth

Max_Amount Real Maximum amount of transactions within

month

Year_Month, Customer_ID, Account_ID, Transaction_Code, and Transaction_Channel are the primary keys
(PKs) of this table.

A dictionary is a typical example of a lookup table, where the key is a word
and the value is the dictionary definition of that word. Similarly, if the lookup table
records the marital status of a customer, the keys might be the marital status codes,
for instance (1, 2, 3), and the values might be the corresponding descriptions, for
instance (single, married, divorced).
Obviously, in a well-designed data mart many different lookup tables are
required. An indicative list of lookup tables useful for the retail banking data mart
is presented in the next sections.

Product Codes

Although the structure of the proposed data mart tables permits the extraction of
information at a product code level, most data mining tasks would typically require
less detailed information, such as data summaries at a product group/subgroup level.
A lookup table (such as the ''L_Product_Code'' table presented in Table 4.6)
can be used to define and store a multilevel grouping hierarchy of product codes,
mapping each product to the corresponding product groups and subgroups. When