The Mining Data Mart - Data Mining Techniques in CRM: Inside Customer Segmentation - page 183

Table 4.32 Purchases by product groups.

Field name Field type Field description

Year_Month Int The number that indicates the year and

month of data

Customer_ID Int Customer ID or loyalty card ID

Item_Group Int UPC group: the highest level of the prod-

uct hierarchy for which summarized
data are kept

Num_of_Items Int Total number of items purchased

Total_Rated_Amount Real The total rated amount

Total_Discount_Amount Real The total discount amount

Year_Month, Customer_ID, and Item_Group are the primary keys (PKs) of this table.

table should correspond to the highest hierarchy level that can address the most
frequent analytical needs. Nevertheless, as mentioned above, apart from pre-
aggregated purchase data, we can always revisit the raw transactional data and the
relevant lookup table to extract information at a lower product level.

LOOKUP INFORMATION

A well-organized lookup table is necessary to classify the product codes in a
multilevel product hierarchy.

The Product Hierarchy

The next table (''L_Items''), Table 4.33, defines a multilevel product hierarchy
that can be used to describe and categorize the items into product groups and
subgroups according to their type. The product hierarchy lookup table can be
combined with the point-of-sale transaction records to enable summarization at
the selected product level, providingmore detailed information than the aggregated
monthly tables. The hierarchy level to be used depends on the specific needs of
the organization. Additional summarizations, for example, by brand (brand labels,
private labels, etc.), are also feasible.
An indicative list of product groups at a high level in the product hierarchy is
presented in Table 4.34.