There are two general types of proof-of-performance projects.
Undirected
Directed
These projects can be tailored to the needs of a particular CDI client.
These projects extract knowledge from the client’s data to demonstrate that
data mining can support meeting the client’s business objectives.
Through such projects the client will be able to develop hands-on expertise in
KDD and will learn how to best invest in KDD to meet their business objectives.
Figure 1: General Types of Proof-of Performance Projects
Undirected Proof-of-Performance Projects
Learn about data mining and data mining tools through a
hands-on analysis of the client’s data.
Observe what types of models the data mining tools generate
in order to intuitively form a judgment as to whether potentially interesting patterns
and relationships can be extracted from the client’s data.
Providing information to the management of a company to
assist in determining whether KDD should be invested in by the business.
Directed Proof-of-Performance Projects
Specific business objectives are predefined
Hypothesis Testing
Develop models from data mining tools to test whether
specific hypotheses about the business are true or false.
Known business objectives
Determine whether significant and/or previously unknown
relationships or patterns exist in the data to improve meeting particular business objectives.
Either project provides hands-on experience with data mining
tools through an analysis of the client’s data.
Either project extracts knowledge from corporate data in
order to make a quantitative judgement as to whether the patterns and relationships exist
in the data better permit the client to meet known business objectives.
Either project provides initial information to the
management of a company to assist in determining how to best invest in KDD and to deploy
KDD into the company’s decision support systems.
The CDI Proof-of-Performance Process
The CDI follows an eight-step process for conducting
proof-of-performance projects.
The CDI works with the company in stages one and two.
The CDI is primarily responsible for stages three and four.
The CDI guides the client in stages five through eight.
Figure 2: The CDI Proof-of-Performance Project Process
It is also strongly emphasized that these projects
assume that the client’s data is in a form, or almost in a form, ready for data
mining. If not, a potentially extensive, customized, data quality analysis project will be
required to analyze, cleanse and format the client’s data to permit
effective data
mining to occur.
