readme

数据挖掘中的一算法 ines算法 c下实现的。适合初学习数据挖掘者借鉴

ines - Induction of NEtwork Structures
(c) 1995-2002 Christian Borgelt

This file provides some explanations on how to use the programs gendb,
ines, and neval. However, it does not explain all options of these
programs. For a list of options, call the programs without any
arguments.

Enjoy,
Christian Borgelt

e-mail: [e-mail]
WWW:    http://fuzzy.cs.uni-magdeburg.de/~borgelt

------------------------------------------------------------------------

Example for using the programs gendb, dom, and ines:

Calling the program gendb with

  gendb t002.pd t002.tab

generates a file t002.tab with 1000 tuples sampled randomly from the
distribution in t002.pd. From this table the field domains can be
determined with the help of the program dom (contained in the table
package; the option -s sorts the domains)

  dom -s t002.tab t002.dom

Now with t002.tab and t002.dom the input files for the ines program
are available. Calling it with

  ines t002.dom t002.tab t002.net -stopord -ebdm

an network structure is induced using a greedy selection of conditions
on the topological order of the attributes (the order in which they
appear in t002.dom) based on the K2 metric (a special case of the
Bayesian Dirichlet metric bdm). This network structure should be the
same as the one given in t002.pd with the exception of some (slight)
deviations in the probabilities.

If one uses xor.pd instead of t002.pd a drawback of the greedy condition
selection used becomes obvious: an exclusiv-or cannot be recognized,
since no attribute alone leads to an improvement.