Ultimately,
the ability to merge a
variety of technologies
with a thorough understanding
of the business is essential
to driving meaningful
results from any data
mining initiative. Familiarity
with a variety of tools,
the different algorithms
they use, and what those
algorithms are capable
of telling someone, is
the key. The steps taken
during the data mining
process are often compared
to those for data warehousing.
Our approach
to data mining is based
on a proven methodology
that blends the underlying
business rules with the
data that is extracted
from the business systems.
The process is outlined
below:
- Knowledge of the
business and the data
generated by that
business comes first
in the process. The
business rules and
data sources are documented
and reviewed.
- The second step
is integrating the
data into one system
or file structure.
Typically this is
called data preparation,
or the ETL (Extract
Transform and Load)
process.
- The third step is
to select the modeling
techniques. Different
tools use different
modeling methods,
and familiarity with
those tools is essential
to getting the desired
results.
- Fourth, the results
of the model must
then be evaluated,
and a decision to
change algorithms
or continue with the
chosen ones must be
made. This feedback
loop is critical.
The sensitivity of
the data and results
always drives the
degree to which the
model is tuned.
- Finally, the fifth
step is the delivery
of the model results
to the business users.
How this information
is displayed can vary
widely. We can use
delivery mechanisms
ranging from simple
spreadsheets to seamless
integration into existing
business intelligence
systems.
Techtrend
has experience in all
aspects of knowledge discovery
and data mining. We provide
our clients the ability
to uncover the hidden
patterns in the data that
they otherwise would never
see. |
