[MUD-Dev] AutoClass

[Hans-Henrik Staerfeldt]

On Sat, 22 Jul 2000, Ling Lo wrote:
>
> Oook, found something interesting, esp for those of you that have audit=
ing
> facilities in your mud:
>=20
> http://ic-www.arc.nasa.gov/ic/projects/bayes-group/autoclass/
>=20
> ---<cut>---
> In previous years, the Bayes group at Ames Research Center developed th=
e
> basic theory and associated algorithms for various kinds of general dat=
a
> analysis techniques. Our earliest efforts were applied to the problem o=
f
> automatic classification of data. We implemented this theory in the
> Autoclass series of programs. AutoClass takes a database of cases
> described by a combination of real and discrete valued attributes, and
> automatically finds the natural classes in that data.=20
[..snip..]
> ---<cut>---
>=20
> This one looks like a tried and tested stylee.

As always you can (should?) guide or control this by having the data you
represent be as close to the problem domain as possible. By problem domai=
n,=20
meaning the sort of information you want to get out of the data. While
the method will make nice clusters, it will not easily tell you what
makes the clusters what they are (i.e. it won't 'name' them). So if you
just throw in all your data, you will have problems knowing what it actua=
lly
is you get out. As usual there is the problem of how to represent=20
time-series data and static data of the same processes, which might requi=
re
some work if you need that analyzed. Otherwise i have seen it work nicely=
 in=20
DNA & protein sequence analysis.=20

Hans Henrik St=E6rfeldt   |    bombman at diku.dk    | work:  hhs at cbs.dtu.dk=
      |
address:                |___  +45 40383492    __|__       +45 45252425   =
  __|
 Dybendalsvej 74 2. th, | Scientific programmer at Center for Biological =
    |
 2720 Vanl=F8se, Danmark. |  Sequence Analysis, Technical University of D=
enmark|




_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
http://www.kanga.nu/lists/listinfo/mud-dev