Databases Reference
In-Depth Information
BUSINESS:
Motivations for the project
Data mining Modeling Objects (DMMO)
abstraction of:
Data sources, structure,
content, flow
Domain (actors, activities,
resources...)
Data mining techniques
Problem types
Algorithms
<< asess >>
<< represent >>
REQUIREMENTS ASSESMENT:
Data mining goals
Stages (Inputs, outputs, dependencies
,...)
<< arise >>
Risks and contingencies plan
Business Goals Model
Costs and benefits
...
PROJECT PLAN:
<< refine+enrich model >>
Data mining goals
<< determine >>
Stages (Inputs, outputs, dependencies, ...)
Risks and contingencies plan
Costs and benefits
...
Feasible Business Goals Model
Fig. 2.
Project plan definition process
understanding for client and data miner but will also be a tool to determine
potential data mining goals and consequently the basis to project plans.
In Fig. 2, DMMO (Data Mining Modelling Objects) denotes the set of the
compounding elements of the Modelling Language. Elements of the business
domain will be abstracted using DMMO, generating the Business Objective
Model. Together with this intermediate model, a document that we have called
Requirements Assessment will be produced. This is a first approach to project
goals in which special attention is given to critical factors (risks, constraints,
information required, ...) that will depend both on the goals themselves and
on the tools and techniques used to achieve them.
The requirement assessment document will be analyzed jointly with the
client to enrich and refine the previous business goals model. The resulting
model from this analysis is what has been called Feasible Business goals Model
in the figure. In this model, goals previously identified but analyzed as not
feasible have been not included, although not removed as these are conclusion
of the projet by themselves. From the refined model and making use once
again of DMMO the project plan will be produced almost in a automatic
way as the model will represent relevant aspects both of the domain and of
the tools themselves. Due to the assumed abstraction capability of DMMO,
the plan will contain detail information about: techniques, tools, kind of data
mining to be solved, inputs, outputs, flow of data and dependencies. Thus,
risk and contingencies, cost, milestones, will be identified.
