Databases Reference
In-Depth Information
be, a project whose requirements are poorly specified will end up with a dis-
appointed end user [26]. In a data mining project the most critical factor is
related to the clear understanding of the business goals.
Both the client and the data miner play an important role when estab-
lishing business goals. The client has to formulate his problem while the data
miner tries to understand it in order to be able to translate it into data min-
ing functions. During this task, it is relevant to keep in mind the following
aspect (inspired from software engineering [26]): “Business domain as well as
functional domain of the problem has to be represented and understood”.
Deeply analyzing any activity of the organization (even external to it)
that generate data that will be potentially used as input in a data mining
project as well as the data themselves and the data mining functions will
highlight important concepts that are common in any data mining project
independently of the domain. However, eliciting, analyzing and graphically
depicting concepts is no easy task [11] and must be developed based on a
systematic method that provide all those elements that make this process less
risky but more controlled.
Hence, first of all it is necessary to set the basis for a definition of elements
that will make it possible to represent or abstract the business domain that
is the target of the project. The goal of this abstraction is to provide a data
mining project manager with a method to systematically describe the goals of
the project. Moreover, once the goals are understood, they can be translated
into data mining goals and then, into data mining problem types. However,
not only identifying the data mining problems to be solved is enough. We
should also be able to find out if the available data to be analyzed fulfil a
set of general requirements or conditions. Every problem type will require
different kinds of data. In the following, we will describe the different existing
issues as well as their requirements.
Data Requirements Assessment
Every data mining project can be collectively described as data analysis and
knowledge extraction to obtain the intelligence of the business. Data do not
only represent the activities or business processes that have generated them
but implicitly carry important hidden knowledge about the business. Extract-
ing this knowledge means making decisions in an intelligent and successful
way.
Though talking about intelligence, data mining does not involve deductive
processes. On the contrary, it is an inductive process that analyzes the data to
extract knowledge: it accepts data from different sources, manipulates them
and obtains an output and patterns of knowledge, that if of good quality, will
be deployed. This is the general setting of the process no matter the domain
or organization we are dealing with.
In the process of data analysis and knowledge extraction, different per-
spectives of the data being analyzed are taken into account: data sources,
