The Mining Data Mart - Data Mining Techniques in CRM: Inside Customer Segmentation - page 149

data mart should be enriched with derived attributes and informative KPIs to
summarize all aspects of the customer's relationship with the organization.
The MCIF typically includes data at a customer level, though the structure
of the data mart presented above also supports the building of a relevant file at a
different level, such as at a product account level for instance. It should be updated
on a regular basis to provide the most recent view of each customer and it should
be stored to track the customer view over time. The main idea is to compose a
good starting point for the rapid creation of an input file for all key future analytical
applications.

Creating the MCIF through Data Processing

The MCIF can be built and maintained by using:

• Database views created in the existing relational database management system,
such as ORACLE, SQL Server, etc.
• Procedures created using data mining tools such as IBM SPSS Modeler
(formerly Clementine) streams.

Nevertheless, the consolidation of all the available information in a single
table is a challenging and demanding job which usually requires complicated data
processing in order to select, transform, and integrate all relevant data. Some
of the most useful data management operations and the corresponding database
functions typically applied for the construction of the MCIF are as follows:

• Select: Used to select or discard a subset of records from a data file based on a
logical condition. In SQL this is expressed as: select < fields > from < data file >

>
• Merge/join: Matches records and adds fields from multiple data sources using
key fields (for instance, the customer ID field). Several types of joins are
available, including inner join, full outer join, partial outer join, and anti-join.
• Aggregate/group by: Aggregates multiple records based on a key field and
replaces the input records with summarized output records. Several summary
statistics can be applied, such as: sum, mean, min, max, standard deviation. In
SQL this is expressed as: select

<

where condition

< key fields > , < summarized fields > from < data

>
• Insert/append: Insert is used to concatenate sets of records. It is useful when
the aim is to combine data sources with the same structure (fields) but with
different records. In SQL this is expressed as: insert into

file > group by < key fields

< > < >

data file , fields

>
• Update: Update is used for filling specific fields with calculated values. It can
be applied to all values of a field or to the values that satisfy a logical condition.

select < fields > from < source data file