Package cascading.pattern.model

Examples of cascading.pattern.model.ModelSchema

54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
  public RandomForestFunction( RandomForestSpec randomForestSpec )
    {
    super( randomForestSpec );

    ModelSchema modelSchema = randomForestSpec.getModelSchema();

    String predictedFieldName = modelSchema.getPredictedFieldNames().get( 0 );

    List<String> predictedCategories = modelSchema.getPredictedCategories( predictedFieldName );

    if( modelSchema.isIncludePredictedCategories() && predictedCategories.isEmpty() )
      throw new IllegalArgumentException( "no predicted categories were set, but include predicted is true" );

    List<String> nodeCategories = randomForestSpec.getModelCategories();
    Sets.SetView<String> difference = Sets.difference( newHashSet( predictedCategories ), newHashSet( nodeCategories ) );
View Full Code Here
556
557
558
559
560
561
562
563
564
565
566
    return tail;
    }

  private Pipe handleTreeModel( Pipe tail, TreeModel model )
    {
    ModelSchema modelSchema = createModelSchema( model );

    Tree tree = createTree( model, modelSchema );

    TreeSpec treeSpec = new TreeSpec( modelSchema, tree );
View Full Code Here
567
568
569
570
571
572
573
574
575
576
577
    return create( tail, modelSchema, new TreeFunction( treeSpec ) );
    }

  private Pipe handleGeneralRegressionModel( Pipe tail, GeneralRegressionModel model )
    {
    ModelSchema modelSchema = createModelSchema( model );

    Set<String> parameterList = GeneralRegressionUtil.createParameters( model );
    Set<String> covariateList = GeneralRegressionUtil.createCovariates( model );
    Set<String> factorsList = GeneralRegressionUtil.createFactors( model );
View Full Code Here
595
596
597
598
599
600
601
602
603
604
605
606
607
    throw new UnsupportedOperationException( "unsupported mining type, got: " + model.getFunctionName() );
    }

  private Pipe handleCategoricalRegressionModel( Pipe tail, RegressionModel model )
    {
    ModelSchema modelSchema = createModelSchema( model );

    List<String> predictedCategories = new ArrayList<String>( modelSchema.getPredictedCategories( modelSchema.getPredictedFieldNames().get( 0 ) ) );

    if( predictedCategories.isEmpty() )
      throw new PatternException( "no categories specified" );

    GeneralRegressionSpec regressionSpec = new GeneralRegressionSpec( modelSchema );
View Full Code Here
617
618
619
620
621
622
623
624
625
626
627
  private Pipe handlePredictionRegressionModel( Pipe tail, RegressionModel model )
    {
    if( model.getRegressionTables().size() != 1 )
      throw new UnsupportedOperationException( "regression model only supports a single regression table, got: " + model.getRegressionTables().size() );

    ModelSchema modelSchema = createModelSchema( model );

    org.dmg.pmml.RegressionTable regressionTable = model.getRegressionTables().get( 0 );

    GeneralRegressionSpec regressionSpec = new GeneralRegressionSpec( modelSchema );
View Full Code Here
632
633
634
635
636
637
638
639
640
641
642
    return create( tail, modelSchema, new PredictionRegressionFunction( regressionSpec ) );
    }

  private Pipe handleClusteringModel( Pipe tail, ClusteringModel model )
    {
    ModelSchema modelSchema = createModelSchema( model );

    if( model.getModelClass() != ClusteringModel.ModelClass.CENTER_BASED )
      throw new UnsupportedOperationException( "unsupported model class, got: " + model.getModelClass() );

    ComparisonMeasure comparisonMeasure = model.getComparisonMeasure();
View Full Code Here
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
    return create( tail, modelSchema, new ClusteringFunction( clusteringSpec ) );
    }

  private Pipe handleMiningModel( Pipe tail, MiningModel model )
    {
    ModelSchema modelSchema = createModelSchema( model );
    List<TreeSpec> models = new LinkedList<TreeSpec>();

    for( Segment segment : model.getSegmentation().getSegments() )
      {
      if( !segment.getPredicate().getClass().equals( True.class ) )
        throw new PatternException( "segment predicates currently not supported, got: " + segment.getPredicate() );

      Model segmentModel = segment.getModel();

      if( segmentModel instanceof TreeModel )
        models.add( new TreeSpec( modelSchema, createTree( (TreeModel) segmentModel, modelSchema ) ) );
      else
        throw new PatternException( "ensemble model currently not supported, got: " + segmentModel );
      }

    EnsembleSpec<TreeSpec> miningSpec = new EnsembleSpec<TreeSpec>( modelSchema, models );

    LOG.debug( "creating: {}, input: {}, output: {}", new Object[]{miningSpec, modelSchema.getInputFields(),
                                                                   modelSchema.getDeclaredFields()} );

    SelectionStrategy strategy = PMMLUtil.getSelectionStrategy( model );

    miningSpec.setSelectionStrategy( strategy );
View Full Code Here
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
    return new Each( tail, inputFields, function, Fields.ALL );
    }

  private ModelSchema createModelSchema( Model model )
    {
    ModelSchema modelSchema = new ModelSchema();

    for( MiningField miningField : model.getMiningSchema().getMiningFields() )
      {
      DataField dataField = getPMMLModel().getDataField( miningField.getName() );

      if( miningField.getUsageType() == FieldUsageType.ACTIVE )
        modelSchema.addExpectedField( createDataFields( dataField ) );
      else if( miningField.getUsageType() == FieldUsageType.PREDICTED )
        modelSchema.setPredictedFields( createDataFields( dataField ) );
      }

    if( modelSchema.getPredictedFieldNames().isEmpty() && defaultPredictedField == null )
      throw new PatternException( "no predicted field name provided in PMML model, use setDefaultPredictedField() method" );

    if( modelSchema.getPredictedFieldNames().isEmpty() )
      modelSchema.setPredictedFields( new ContinuousDataField( defaultPredictedField ) );

    return modelSchema;
    }
View Full Code Here
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
      throw new IllegalArgumentException( "given selection strategy must support parallel models, got: " + ensembleSpec.getSelectionStrategy() );

    if( ensembleSpec.getModelSpecs().size() < 2 )
      throw new IllegalArgumentException( "ensembles must have more than 1 model" );

    ModelSchema modelSchema = ensembleSpec.getModelSchema();

    Fields predictedFields = modelSchema.getPredictedFields();
    Fields keyFields = modelSchema.getKeyFields();

    if( keyFields.isNone() )
      {
      keyFields = new Fields( "ensemble-primary-key", String.class );
      pipe = new InsertGUID( pipe, keyFields );
      }

    boolean isCategorical = ensembleSpec.isPredictedCategorical();

    // the parallel bits
    List<Pipe> pipes = new ArrayList<Pipe>();

    for( int i = 0; i < ensembleSpec.getModelSpecs().size(); i++ )
      {
      Spec spec = (Spec) ensembleSpec.getModelSpecs().get( i );

      if( spec instanceof TreeSpec )
        pipes.add( createScoringPipe( i, pipe, modelSchema, new TreeFunction( (TreeSpec) spec, isCategorical, false ) ) );
      }

    pipe = new GroupBy( "vote", pipes.toArray( new Pipe[ pipes.size() ] ), keyFields );

    SelectionBuffer buffer;

    if( isCategorical )
      buffer = new CategoricalSelectionBuffer( ensembleSpec );
    else
      buffer = new PredictionSelectionBuffer( ensembleSpec );

    pipe = new Every( pipe, predictedFields, buffer, Fields.SWAP );

    if( modelSchema.getKeyFields().isNone() )
      pipe = new Discard( pipe, keyFields );

    setTails( pipe );
    }
View Full Code Here
46
47
48
49
50
51
52
53
54
55
56
57
58
    super( regressionSpec );

    if( regressionSpec.getNormalization() == null )
      throw new IllegalArgumentException( "normalization may not be null" );

    ModelSchema modelSchema = regressionSpec.getModelSchema();

    DataField predictedField = modelSchema.getPredictedField( modelSchema.getPredictedFieldNames().get( 0 ) );

    if( !( predictedField instanceof CategoricalDataField ) )
      throw new IllegalArgumentException( "predicted field must be categorical" );

    if( ( (CategoricalDataField) predictedField ).getCategories().size() != regressionSpec.getRegressionTables().size() )
View Full Code Here
TOP

Related Classes of cascading.pattern.model.ModelSchema

  • cascading.pattern.datafield.CategoricalDataField
  • cascading.pattern.datafield.ContinuousDataField
  • cascading.pattern.datafield.DataField
  • cascading.pattern.ensemble.ParallelEnsembleAssembly
  • cascading.pattern.ensemble.SimpleEnsemblePlatformTest
  • cascading.pattern.function.RandomForestFunction
  • cascading.pattern.model.generalregression.CategoricalRegressionFunction
  • cascading.pattern.pmml.PMMLPlanner
  • cascading.tuple.Fields
    • Copyright © 2018 www.massapicom. All rights reserved.
    All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.