Java Reference
In-Depth Information
Table 9-12
javax.datamining.supervised.classificaton.ClassificationModel
interfaces (continued)
Methods
Description
getApplicationName(): String
Returns the name of the application that generated this
model.
getVersion(): String
Returns the JDM version by which the model was built.
getMajorVersion(): String
Returns the major version of JDM by which the model was
built.
getMinorVersion(): String
Returns the minor version of JDM by which the model was
built.
getProviderName(): String
Returns the name of the JDM provider, i.e., the vendor, that
built the model.
getProviderVersion(): String
Returns the version number of the JDM system provided by
a vendor that was used to build the model.
produced by different classification algorithms. Lines 16 to 26 show
retrieval of the
TreeModelDetail
object that encapsulates the decision
tree nodes; using its methods, applications can display the decision
tree. The
TreeNode
encapsulates a node's prediction value, the num-
ber of cases per node, the rule associated with the node, and the
node's child and parent node references.
In this example, we retrieve the first level node details. Lines 27
to 37 show retrieval of the
NaiveBayesModelDetail
object that encap-
sulates the target probability and pair probability details. We
retrieve the target probability associated with the target
attrite
attribute value
Attriter
, pair probabilities associated with the
capital
gains
values, and the attrite value
Non-attriter
. Using these details
applications can show the target and attribute value correlations.
Lines 37 to 43 show the retrieval of the
SVMClassificationModelDetail
object that encapsulates coefficients associated with the attribute
and target value pairs. In line 40, we retrieve the coefficients associ-
ated with the
capital gains
attribute values and the
attrite
value
Attriter
. Lines 45 and 46 show the retrieval of the
NeuralNetwork-
ModelDetail
object that encapsulates the details of the weights associ-
ated with the neurons. In lines 47 to 58, we retrieve the weights
associated with each layer and the parent and child neuron possible
combinations.
Search WWH ::
Custom Search