Databases Reference
In-Depth Information
major difference is that in the regression task the attribute to be predicted
is continuous i.e., it can take on any real valued number or any integer
number in an arbitrarily large range rather than discrete.
1.2.1.7.
Summarization
This involves methods for finding a compact description for a subset of data.
A simple example would be tabulating the mean and standard deviations
for all attributes. In other words, the aim of the summarization task is to
produce a characteristic description of each class of tuples in the target
dataset.
19
This kind of description somehow summarizes the attribute
values of the tuples that belong to a given class. That is, each class
description can be regarded as a conjunction of some properties shared
by all (or most) tuples belonging to the corresponding class.
The discovered class descriptions can be expressed in the form of
“IF-THEN” rules, interpreted as follows: “if a tuple belongs to the class
indicated in the antecedent of the rule, then the tuple has all the properties
mentioned in the consequent of the rule”. It should be noticed that in
summarization rules the class is specified in the antecedent (“if part”) of
the rule, while in classification rules the class is specified in the consequent
(“then part”) of the rule.
1.2.1.8.
Causation Modeling
This task involves the discovery of relationships of cause and effect among
attributes. Causal rules are also “if-then” rules, like dependence rules, but
causal rules are intuitively stronger than dependence rules.
1.3. Intelligent Agents
1.3.1.
Evolutionary computing
This section provides an overview of biologically inspired algorithm
drawn from an evolutionary metaphor.
20,21
In biological evolution,
species are positively or negatively selected depending on their relative
success in surviving and reproducing in their current environment.
Differential survival and variety generation during reproduction provide
the engine for evolution. These concepts have metaphorically inspired a
family of algorithms known as evolutionary computation. The algorithms
like
genetic
algorithms,
genetic
programming,
evolution
strategies,
