Customer Segmentation - Data Mining Techniques in CRM: Inside Customer Segmentation - page 205

groups of telephone lines (MSISDNs in mobile telephony), and so on. The
selection of the appropriate segmentation level depends on the subsequent
marketing activities that the segments are about to support. It also determines
the aggregation level of the modeling dataset that is going to be constructed.
Data Understanding, Preparation, and Enrichment
The investigation and assessment of the available data sources is followed by
data acquisition, integration, and processing for the needs of segmentation
modeling. The data understanding and preparation phase is probably the most
time-consuming phase of the project and includes tasks such as:
1. Data source investigation: The available data sources should be evaluated in
terms of accessibility and validity. This phase also includes initial data collection
and exploration in order to understand the available data.
2. Defining the data to be used: The next step in the procedure involves the
definition of the data to be used for the needs of the analysis.
The selected data should cover all the behavioral dimensions that will be
used for the segmentation as well as all the additional customer information
that will be used to gain deeper insight into the segments.
3. Data integration and aggregation: The initial raw data should be consol-
idated to create the final modeling dataset that will be used for identification
of the segments. This task typically includes the collection, filtering, merg-
ing, and aggregation of the raw data. But first the structure of the modeling
dataset should be defined, including its contents, time frame of used data, and
aggregation level.
For behavioral segmentation applications, a recent ''view'' of the customers'
behavior should be constructed and used. This ''view'' should summarize the
behavior of each customer by using at least sixmonths of recent data (Figure 5.5).
The aggregation level of the modeling dataset should correspond to the
required segmentation level. If the goal, for instance, is to segment bank
customers, then the final dataset should be at a customer level. If the goal is
to segment telephone lines (MSISDNs), the final dataset should be at a line
level. To put it in a simpler way, clustering techniques reveal natural groupings
of records. So no matter where we start from, the goal is the construction of a
final, one-dimensional, flat table, which summarizes behaviors at the selected
analysis level.
This phase concludes with the retrieval and consolidation of data from
multiple data sources (ideally from the organization's mining data mart and/or
MCIF) and the construction of the modeling dataset.