Database Reference
In-Depth Information
// transMeal2
¼
(calories, numberOfGuests, cutlery) with
// cutlery
¼
(knife, fork, spoon, skewer) !skewer is added
// ---- Transforms meal vector from 'meal2' into 'meal' ---
// Create meal 'soup' for spoons, 2 persons, 57$, 19 kcal:
double[] soupArr
¼
{0, 2, 57, 19000};
MiningVector soup
¼
new MiningVector(soupArr);
soup.setMetaData(meal2);
// Transform 'soup' from 'meal2' into 'meal' basis:
MiningVector transSoup
¼
mealOp.transform(soup);
System.out.println("transformed 'soup': " + transSoup);
// transSoup
¼
(19000, 2, 2(
¼
spoon))
12.1.3 The Data Mining Framework
12.1.3.1 Models
The abstract class
MiningModel
represents the data mining model which is mainly
the mining function
.
The central method of
MiningModel
is
applyModelFunction
which takes a mining vector as argument and returns the function value
.
Thus,
applyModelFunction
is used to apply the mining model to data. There exists a
second application method
applyModel
which is more general and returns objects
(e.g., a mining vector for SV clustering, an item set for an association rule model, or
a node for a decision tree model).
Each class representing a type of data mining models extends
MiningModel.
For
instance,
AssociationRulesMiningModel
extends
MiningModel
for association rules
and contains the implementations (
applyModel
for rules and PMML export/import
of rules). For a special association rule model,
AssociationRuleMiningModel
may
be further subclassed. For instance, for flat association rules, it may be useful to
introduce a new class
FlatRulesMiningModel
which extends
AssociationRulesMi-
ningModel.
XELOPES already contains a wide hierarchy of all basic classes of data
mining models including the main implementations like the
apply
methods and
PMML serialization. If the user requires a special model, he/she can extend one of
the existing models.
Because of the wide variety of data mining models and algorithms, a two-level
system of their classification is used.
The
function
level defines the basic types of mining models like
Clustering
and
Regression.
The mining models of XELOPES are organized in packages whose
names correspond to the functions. For instance, all classification models are
contained in the package
Classification
which contains further subpackages for
special classification models.