Re: DM: Datamining Definition...and Machine Learning Definition.

> From: "Franklin Wayne Poley" <culturex@vcn.bc.ca>
> ...
> If we go with such a broad term then data mining/knowledge extraction
> becomes synonymous with machine learning does it not?

Unfortunately, data mining has become whatever the marketing people
trying to sell expensive software define it to be.

Machine learning is traditionally concerned with small, noise-free
data sets, primarily with categorical variables. In recent years
the ML people have shown more interest in noisy data and continuous
variables, but they still seem to view noisy data as an aberration.

Data mining is traditionally concerned with huge, noisy data sets
with all kinds of messy variables. Primarily, the purpose of data
mining is to create a predictive model for a specific target
variable (such as customer purchasing, credit card fraud, etc.) or
to see if there are any predictive relationships among a large
number of variables (e.g., "associations and sequences", market
basket analysis).  Predictive models for noisy data were called
"statistical models" before some marketing person came up with the
term "data mining", which, by the way, used to be a derogatory term
in the statistical literature.  Secondarily, data mining is
concerned with detecting outliers (anomalies, novelties), which is
another application of statistical models.  Ultimately, data mining
is used to make decisions--usually business decisons but perhaps
medical decisions or various other kinds of decisions. Making
decisions based on noisy data is the province of statistical
decision theory, which is used in the SAS Enterprise Miner product.
So I choose to define data mining as the application of statistical
decision theory to huge, messy data sets to maximize profits.

-- 

Warren S. Sarle       SAS Institute Inc.   The opinions expressed here
saswss@unx.sas.com    SAS Campus Drive     are mine and not necessarily
(919) 677-8000        Cary, NC 27513, USA  those of SAS Institute.

• References:
  • Re: DM: Datamining Definition...and Machine Learning Definition.
    • From: Franklin Wayne Poley

• Prev by Date: DM: New PhD seeks DFW data mining/statistical work
• Next by Date: Re: DM: Datamining Definition & UEA MSc & chaid question
• Prev by thread: DM: New PhD seeks DFW data mining/statistical work
• Next by thread: DM: CEC2000 - Challenge and Competitions
• Index(es):
  • Date
  • Thread