weka.core.pmml

Class MiningSchema

java.lang.Object

weka.core.pmml.MiningSchema

All Implemented Interfaces:

java.io.Serializable

public class MiningSchema
extends java.lang.Object
implements java.io.Serializable

This class encapsulates the mining schema from a PMML xml file. Specifically, it contains the fields used in the PMML model as an Instances object (just the header). It also contains meta information such as value ranges and how to handle missing values, outliers etc. We also store various other PMML elements here, such as the TransformationDictionary, DerivedFields and Targets (if defined). They are not part of the mining schema per se, but relate to inputs used by the model and it is convenient to store them here.

Version:

$Revision: 5562 $

Author:

Mark Hall (mhall{[at]}pentaho{[dot]}com)

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
MiningSchema(org.w3c.dom.Element model, Instances dataDictionary, weka.core.pmml.TransformationDictionary transDict) Constructor for MiningSchema.

Method Summary

Modifier and Type	Method and Description
void	applyMissingAndOutlierTreatments(double[] values) Apply both missing and outlier treatments to an incoming instance.
void	applyMissingValuesTreatment(double[] values) Apply the missing value treatments (if any) to an incoming instance.
void	applyOutlierTreatment(double[] values) Apply the outlier treatment methods (if any) to an incoming instance.
void	convertNumericAttToNominal(int index, java.util.ArrayList<java.lang.String> newVals) Convert a numeric attribute in the mining schema to nominal.
void	convertStringAttsToNominal() Method to convert any string attributes in the mining schema Instances to nominal attributes.
java.util.ArrayList<DerivedFieldMetaInfo>	getDerivedFields()
Instances	getFieldsAsInstances() Get the all the fields (both mining schema and derived) as Instances.
java.util.ArrayList<MiningFieldMetaInfo>	getMiningFields()
Instances	getMiningSchemaAsInstances() Get the mining schema fields as an Instances object.
TargetMetaInfo	getTargetMetaData() Get the Target meta data.
weka.core.pmml.TransformationDictionary	getTransformationDictionary() Get the transformation dictionary .
boolean	hasTargetMetaData() Returns true if there is Target meta data.
java.lang.String	toString() Get a textual description of the mining schema.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

MiningSchema

public MiningSchema(org.w3c.dom.Element model,
                    Instances dataDictionary,
                    weka.core.pmml.TransformationDictionary transDict)
             throws java.lang.Exception

Constructor for MiningSchema.

Parameters:

model - the Element encapsulating the pmml model

dataDictionary - the data dictionary as an Instances object

Throws:

java.lang.Exception - if something goes wrong during construction of the mining schema

Method Detail

applyMissingValuesTreatment

public void applyMissingValuesTreatment(double[] values)
                                 throws java.lang.Exception

Apply the missing value treatments (if any) to an incoming instance.

Parameters:

values - an array of doubles in order of the fields in the mining schema that represents the incoming instance (note: use PMMLUtils.instanceToSchema() to generate this).

Throws:

java.lang.Exception - if something goes wrong during missing value handling

applyOutlierTreatment

public void applyOutlierTreatment(double[] values)
                           throws java.lang.Exception

Apply the outlier treatment methods (if any) to an incoming instance.

Parameters:

values - an array of doubles in order of the fields in the mining schema that represents the incoming instance (note: use PMMLUtils.instanceToSchema() to generate this).

Throws:

java.lang.Exception - if something goes wrong during outlier treatment handling

applyMissingAndOutlierTreatments

public void applyMissingAndOutlierTreatments(double[] values)
                                      throws java.lang.Exception

Apply both missing and outlier treatments to an incoming instance.

Parameters:

values - an array of doubles in order of the fields in the mining schema that represents the incoming instance (note: use MappingInfo.instanceToSchema() to generate this).

Throws:

java.lang.Exception - if something goes wrong during this process

getFieldsAsInstances

public Instances getFieldsAsInstances()

Get the all the fields (both mining schema and derived) as Instances. Attributes are in order of those in the mining schema, followed by derived attributes from the TransformationDictionary followed by derived attributes from LocalTransformations.

Returns:

all the fields as an Instances object

getMiningSchemaAsInstances

public Instances getMiningSchemaAsInstances()

Get the mining schema fields as an Instances object.

Returns:

the mining schema fields as an Instances object.

getTransformationDictionary

public weka.core.pmml.TransformationDictionary getTransformationDictionary()

Get the transformation dictionary .

Returns:

the transformation dictionary or null if none is defined.

hasTargetMetaData

public boolean hasTargetMetaData()

Returns true if there is Target meta data.

Returns:

true if there is Target meta data

getTargetMetaData

public TargetMetaInfo getTargetMetaData()

Get the Target meta data.

Returns:

the Target meta data

convertStringAttsToNominal

public void convertStringAttsToNominal()

Method to convert any string attributes in the mining schema Instances to nominal attributes. This may be necessary if there are no Value elements defined for categorical fields in the data dictionary. In this case, elements in the actual model definition will probably reveal the valid values for categorical fields.

convertNumericAttToNominal

public void convertNumericAttToNominal(int index,
                                       java.util.ArrayList<java.lang.String> newVals)

Convert a numeric attribute in the mining schema to nominal.

Parameters:

index - the index of the attribute to convert

newVals - an ArrayList of the values of the nominal attribute

getDerivedFields

public java.util.ArrayList<DerivedFieldMetaInfo> getDerivedFields()

getMiningFields

public java.util.ArrayList<MiningFieldMetaInfo> getMiningFields()

toString

public java.lang.String toString()

Get a textual description of the mining schema.

Overrides:

toString in class java.lang.Object

Returns:

a textual description of the mining schema