Examples of cascading.pattern.model.generalregression.RegressionTable

Package cascading.pattern.model.generalregression

Examples of cascading.pattern.model.generalregression.RegressionTable

cascading.pattern.model.generalregression.RegressionTable
Class RegressionTable simply holds a set of {@link Parameter} instances.
If used for classification or categorization with CategoricalRegressionFunction, the table must have a {@code targetCategory} value.


    Set<String> parameterList = GeneralRegressionUtil.createParameters( model );
    Set<String> covariateList = GeneralRegressionUtil.createCovariates( model );
    Set<String> factorsList = GeneralRegressionUtil.createFactors( model );


    RegressionTable regressionTable = GeneralRegressionUtil.createPPMatrix( model, parameterList, factorsList, covariateList );


    LinkFunction linkFunction = LinkFunction.getFunction( model.getLinkFunction().value() );


    GeneralRegressionSpec modelParam = new GeneralRegressionSpec( modelSchema, regressionTable, linkFunction );

View Full Code Here

 */
class RegressionUtil
  {
  public static RegressionTable createTable( org.dmg.pmml.RegressionTable regressionTable )
    {
    RegressionTable generalRegressionTable = new RegressionTable();


    String targetCategory = regressionTable.getTargetCategory();


    if( targetCategory != null )
      generalRegressionTable.setTargetCategory( targetCategory );


    generalRegressionTable.addParameter( new Parameter( "intercept", regressionTable.getIntercept() ) );


    int count = 0;


    for( CategoricalPredictor predictor : regressionTable.getCategoricalPredictors() )
      {
      String name = predictor.getName().getValue();
      String value = predictor.getValue();
      double coefficient = predictor.getCoefficient();


      generalRegressionTable.addParameter( new Parameter( "f" + count++, coefficient, new FactorPredictor( name, value ) ) );
      }


    for( NumericPredictor predictor : regressionTable.getNumericPredictors() )
      {
      String name = predictor.getName().getValue();
      int exponent = predictor.getExponent();


      double coefficient = predictor.getCoefficient();


      generalRegressionTable.addParameter( new Parameter( "f" + count++, coefficient, new CovariantPredictor( name, exponent ) ) );
      }


    return generalRegressionTable;
    }

View Full Code Here

  {
  private static final Logger LOG = LoggerFactory.getLogger( GeneralRegressionUtil.class );


  public static RegressionTable createPPMatrix( GeneralRegressionModel model, Set<String> parameterList, Set<String> factorsList, Set<String> covariateList )
    {
    RegressionTable regressionTable = new RegressionTable();


    for( org.dmg.pmml.PCell modelPCell : model.getParamMatrix().getPCells() )
      {
      String parameterName = modelPCell.getParameterName();
      double beta = modelPCell.getBeta();
      Integer df = modelPCell.getDf();


      regressionTable.addParameter( new cascading.pattern.model.generalregression.Parameter( parameterName, beta, df.intValue() ) );
      }


    for( org.dmg.pmml.PPCell modelPPCell : model.getPPMatrix().getPPCells() )
      {
      String parameterName = modelPPCell.getParameterName();
      String predictorName = modelPPCell.getPredictorName().getValue();
      String value = modelPPCell.getValue();


      cascading.pattern.model.generalregression.predictor.Predictor predictor;


      if( factorsList.contains( predictorName ) )
        predictor = new FactorPredictor( predictorName, value );
      else if( covariateList.contains( predictorName ) )
        predictor = new CovariantPredictor( predictorName, Long.parseLong( value ) );
      else
        throw new IllegalStateException( "unknown predictor name: " + predictorName );


      regressionTable.getParameter( parameterName ).addPredictor( predictor );
      }


    Set<String> parameterNames = regressionTable.getParameterNames();


    if( !parameterNames.containsAll( parameterList ) )
      LOG.warn( "different set of parameters: {}", Sets.difference( parameterNames, parameterList ) );


    return regressionTable;

View Full Code Here


    GeneralRegressionSpec regressionSpec = new GeneralRegressionSpec( modelSchema );


    regressionSpec.setLinkFunction( LinkFunction.LOGIT );


    RegressionTable table = new RegressionTable();


    table.addParameter( new Parameter( "p0", -16.9456960387809d ) );
    table.addParameter( new Parameter( "p1", 11.7592159418536d, new CovariantPredictor( "sepal_length" ) ) );
    table.addParameter( new Parameter( "p2", 7.84157781514097d, new CovariantPredictor( "sepal_width" ) ) );
    table.addParameter( new Parameter( "p3", -20.0880078273996d, new CovariantPredictor( "petal_length" ) ) );
    table.addParameter( new Parameter( "p4", -21.6076488529538d, new CovariantPredictor( "petal_width" ) ) );


    regressionSpec.addRegressionTable( table );


    PredictionRegressionFunction regressionFunction = new PredictionRegressionFunction( regressionSpec );

View Full Code Here


    ModelSchema modelSchema = new ModelSchema( expectedFields, predictedFields );


    GeneralRegressionSpec regressionSpec = new GeneralRegressionSpec( modelSchema );


    RegressionTable regressionTable = new RegressionTable();


    regressionTable.addParameter( new Parameter( "intercept", 2.24166872421148d ) );


    regressionTable.addParameter( new Parameter( "p1", 0.53448203205212d, new CovariantPredictor( "sepal_width" ) ) );
    regressionTable.addParameter( new Parameter( "p2", 0.691035562908626d, new CovariantPredictor( "petal_length" ) ) );
    regressionTable.addParameter( new Parameter( "p3", -0.21488157609202d, new CovariantPredictor( "petal_width" ) ) );


    regressionTable.addParameter( new Parameter( "p4", 0d, new FactorPredictor( "species", "setosa" ) ) );
    regressionTable.addParameter( new Parameter( "p5", -0.43150751368126d, new FactorPredictor( "species", "versicolor" ) ) );
    regressionTable.addParameter( new Parameter( "p6", -0.61868924203063d, new FactorPredictor( "species", "virginica" ) ) );


    regressionSpec.addRegressionTable( regressionTable );


    PredictionRegressionFunction regressionFunction = new PredictionRegressionFunction( regressionSpec );

View Full Code Here

    GeneralRegressionSpec regressionSpec = new GeneralRegressionSpec( modelSchema );


    regressionSpec.setNormalization( new SoftMaxNormalization() );


    {
    RegressionTable regressionTable = new RegressionTable( "versicolor" );


    regressionTable.addParameter( new Parameter( "intercept", 86.7061379450354d ) );
    regressionTable.addParameter( new Parameter( "p0", -11.3336819785783d, new CovariantPredictor( "sepal_length" ) ) );
    regressionTable.addParameter( new Parameter( "p1", -40.8601511206805d, new CovariantPredictor( "sepal_width" ) ) );
    regressionTable.addParameter( new Parameter( "p2", 38.439099544679d, new CovariantPredictor( "petal_length" ) ) );
    regressionTable.addParameter( new Parameter( "p3", -12.2920287460217d, new CovariantPredictor( "petal_width" ) ) );


    regressionSpec.addRegressionTable( regressionTable );
    }


    {
    RegressionTable regressionTable = new RegressionTable( "virginica" );


    regressionTable.addParameter( new Parameter( "intercept", -111.666532867146d ) );
    regressionTable.addParameter( new Parameter( "p0", -47.1170644419116d, new CovariantPredictor( "sepal_length" ) ) );
    regressionTable.addParameter( new Parameter( "p1", -51.6805606658275d, new CovariantPredictor( "sepal_width" ) ) );
    regressionTable.addParameter( new Parameter( "p2", 108.27736751831d, new CovariantPredictor( "petal_length" ) ) );
    regressionTable.addParameter( new Parameter( "p3", 54.0277175236148d, new CovariantPredictor( "petal_width" ) ) );


    regressionSpec.addRegressionTable( regressionTable );
    }


    {
    RegressionTable regressionTable = new RegressionTable( "setosa" );


    regressionTable.addParameter( new Parameter( "intercept", 0d ) );


    regressionSpec.addRegressionTable( regressionTable );
    }


    CategoricalRegressionFunction regressionFunction = new CategoricalRegressionFunction( regressionSpec );

View Full Code Here

TOP

Related Classes of cascading.pattern.model.generalregression.RegressionTable

cascading.pattern.model.generalregression.expression.ExpressionEvaluator

cascading.pattern.model.ModelTest

cascading.pattern.pmml.GeneralRegressionUtil

cascading.pattern.pmml.PMMLPlanner

cascading.pattern.pmml.RegressionUtil

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.