A development methodology
specific to data mining is CRISP-DM (CRoss Industry Standard Process for Data
Mining) a methodology conceived in 1996 by IT professionals and was based on
their experiences of data mining implementations (The modelling agency, 2000
p.3).
CRISP DM is broken
into six distinct steps, at some points the outcome of the step may require
repeating the previous step.
Step 1 – Business understanding
This step is broken
into several sub-steps which set the scene for the data mining development, it
is broadly similar to the PID (Project Initiation Document) that makes up a
PRINCE2 project (OGC, 2005 p.40-41).
Determine business objectives
Considers the business
and its overall goals/objectives so as to set the scene (The Modelling Agency,
2000 p.16), discussing data warehouses Mukherjee and D’Souza (2003 p.84) agree
with their statement that “DW
implementation can be considered a success not only because it satisfies a need
at a point in time, but also because it serves the continuing needs of an
organization”.
Assess situation
Looks at available
resources and any associated legal issues/risks (The Modelling Agency, 2000
p.17), this prevents undertaking work that cannot be completed due to
resourcing issues and prevents work being undertaken that cannot be made use of
because of legal issues. The impact of
the development can also be factored in to compare the intended value gained
against the resource required (The Modelling Agency, 2000 p.18).
Determine data mining goals
It is important in
any project to ensure that the aim and associated objectives are clear (OGC,
2005 p.50). This sub-step of business
understanding ensures that the goals are clearly defined and understood (The
Modelling Agency, 2000 p.18), this allows the final outcome to be measured
against these to ensure that the requirements have been met.
Project plan
The final sub-step of
business understanding is the construction of a project plan, breaking the
project into the steps that will be undertaken, the resources that will be
required at each stage and identifies dependencies that may cause bottle necks
in the delivery of the project (The Modelling Agency, 2000 p.19). Traditional project management
techniques/solutions can be made use of in undertaking this such as PRINCE2 and
Microsoft Project.
This comment has been removed by a blog administrator.
ReplyDeleteThis comment has been removed by a blog administrator.
ReplyDeleteGood information you shared. keep posting.
ReplyDeletedata science course in noida