Database Reference
In-Depth Information
12.1.3.3 Mining Settings
MiningSettings
contains the general parameters of a mining model independent
of the specific mining algorithm that has created the model. For instance, an
association rule settings class must contain the minimum support and confidence
parameters because they are required for
each
association rule model. In con-
trast, parameters like the decomposition size, which is required for specific
association rule decomposition algorithms, are contained in the algorithm-
specific parameter class
MiningAlgorithmSpecification
that will be described in
the next section.
MiningSettings
has a reference to its mining model. The most important variable
of
MiningSettings
is
dataSpecification
of the class
MiningDataSpecification.
It
contains the metadata of the training data used to build the model and is referred
to as the
metadata of the mining model
.
MiningSettings
contains a
verifySettings
method which checks all parameters of
the settings class for correctness and completeness.
Similar to
MiningModel
, each class representing a type of data mining settings
extends
MiningSettings.
For the example at the beginning of this section, the
settings of association rule models are contained in the class
AssociationRule-
Settings
which extends
MiningSettings.
Of course, this is the settings class associ-
ated with
AssociationRuleMiningModel
mentioned in the previous section.
Along with all mining models, XELOPES provides their associated settings
classes containing all basic parameters of the respective models.
MiningSettings
contains the same variables
function
and
algorithm
for storing
the function and algorithm type of the mining model. Their values are identical to
those of the associated mining model.
12.1.3.4 Mining Algorithm Specification
The algorithm-specific class
MiningAlgorithmSpecification
contains the function
and algorithm, the name, the class path, the version, and an array of specific
parameters of a mining algorithm. This array contains the specific parameters
defined by the
MiningAlgorithmParameter
class. Every parameter is described
by its name, type, value, description, and setter method and contains the refer-
ence to its associated
MiningAlgorithmSpecification
object.
In most XELOPES distributions, the complete information of
MiningAlgorithm-
Specification
for all algorithms and parameters is stored in the configuration file
algorithms.xml
.
Example 12.8
Example of the section of the fast sequential algorithm
Sequential
of
algorithms.xml
:
<
AlgorithmSpecification name
¼
"Sequential"
function
¼
"Sequential"
algorithm
¼
"sequenceAnalysis"