PMML features and attributes

Using External Models With Vertica gives an overview of the features Vertica supports for working with external models.

Using external models with Vertica gives an overview of the features Vertica supports for working with external models. This topic provides additional details on limitations in how Vertica supports working with PMML models.

In order to be supported by Vertica, PMML models:

Must not contain a data preprocessing step.
Must encode only these model types: k-means, linear regression, logistic regression, random forest (classifier and regressor), XGBoost (classifier and regressor), GeneralRegressionModel, and TreeModel.

Supported PMML tags and attributes

The following table lists supported PMML tags and their attributes.

XML-tag name	Ignored attributes	Supported attributes	Unsupported attributes	Ignored sub-tags	Supported sub-tags	Unsupported sub-tags
Categories	-		-	-	Category	Extension
Category	-	value (required)	-	-	-	Extension
CategoricalPredictor	-	name (required) value (required) coefficient (required)	-	-	-	Extension
Cluster	size	id name	-	KohonenMap Covariances	NUM-ARRAY	Extension Partition
ClusteringField	-	field (required) isCenterField (only "true" is supported) compareFunction	fieldWeight similarityScale	-	-	Extension Comparisons
ClusteringModel	modelName	functionName (required, only "clustering" is supported) algorithmName modelClass (required, only "centerBased" is supported) numberOfClusters(required) isScorable (only "true" is supported)	-	ModelVerification	MiningSchema ComparisonMeasure ClusteringField Cluster	Extension Output ModelStats ModelExplanation LocalTransformations MissingValueWeights ModelVerification
ComparisonMeasure	minimum maximum	kind (required, only "distance" is supported) compareFunction	-	-	euclidean squaredEuclidean	Extension chebychev cityBlock minkowski simpleMatching jaccard tanimoto binarySimilarity
CompoundPredicate	-	booleanOperator (required)	-	-	SimplePredicate CompoundPredicate SimpleSetPredicate True False	Extension
CovariateList	-	-	-	-	Predictor	Extension
DataDictionary	-	numberOfFields	-	-	DataField	Extension Taxonomy
DataField	displayName	name (required) optype (required) dataType (required)	taxonomy isCyclic	-	Value	Extension Interval Value
FactorList	-	-	-	-	Predictor	Extension
False	-	-	-	-	-	Extension
GeneralRegressionModel	modelName targetVariableName startTimeVariable subjectIDVariable	modelType (required) functionName (required) algorithmName targetReferenceCategory cumulativeLink linkFunction linkParameter trialsVariable trialsValue distribution distParameter offsetVariable offsetValue modelDF isScoreable (only "true" is supported)	endTimeVariable statusVariable baselineStrataVariable	ModelVerification PCovMatrix	MiningSchema Output ParameterList PPMatrix ParamMatrix FactorList CovariateList	Extension ModelStats ModelExplanation Targets LocalTransformations EventValues BaseCumHazardTables
Header	copyright description modelVersion	-	-	Extension Application Annotation Timestamp	-	-
MiningField	importance missingValueTreatment	name(required) usageType optype	outliers lowValue highValue missingValueReplacement invalidValueTreatment	-	-	Extension
MiningModel	modelName algorithmName	functionName (required) isScoreable (only "true" is supported)	-	ModelVerification	MiningSchema Output Segmentation Targets	Extension DecisionTree ModelStats ModelExplanation LocalTransformations Regression
MiningSchema	-	-	-	-	MiningField	Extension
Node	-	id score recordCount defaultChild	-	-	SimplePredicate CompoundPredicate SimpleSetPredicate True False ScoreDistribution Node	Extension Partition EmbeddedModel
NumericPredictor	-	name (required) exponent coefficient (required)	-	-	-	Extension
Output	-	-	-	-	OutputField	Extension
OutputField	displayName opType	name (required) dataType (required) feature value isFinalResult	targetField ruleFeature algorithm rankBasis segmentId rank rankOrder isMultiValued	-	-	Extension Decisions Value
Parameter	-	name (required) label	referencePoint	-	-	Extension
ParameterList	-	-	-	-	Parameter	Extension
ParamMatrix	-	-	-	-	PCell	Extension
PCell	-	parameterName (required) targetCategory beta (required) df	-	-	-	Extension
PPCell	-	parameterName (required) predictorName (required) parameterName (required) targetCategory	-	-	-	Extension
PPMatrix	-	-	-	-	PPCell	Extension
PMML	-	version (required) xmlns	-	MiningBuildTask	Header DataDictionary ClusteringModel RegressionModel	TransformationDictionary Extension any unsupported model type
Predictor	-	name (required) contrastMatrixType	-	-	Categories Matrix	Extension
RegressionModel	modelName targetFieldName modelType	functionName (required) algorithmName normalizationMethod isScorable (only "true" is supported)	-	ModelVerification	MiningSchema RegressionTable	Extension Output ModelStats ModelExplanation LocalTransformations Targets ModelVerification
RegressionTable	-	intercept (required) targetCategory	-	-	CategoricalPredictor NumericPredictor	Extension CategoricalPredictor PredictorTerm
Segment	-	id weight		-	PREDICATE MODEL-ELEMENT	Extension VariableWeight
Segmentation	missingPredictionTreatment missingThreshold	multipleModelMethod (required)	-	-	Segment	Extension
SimplePredicate	-	field (required) operator (required) value	-	-	-	Extension
SimpleSetPredicate	-	field (required) booleanOperator (required)	-	-	ARRAY	Extension
Target	-	field optype rescaleConstant rescaleFactor	castInteger min max	-	-	Extension TargetValue
Targets	-	-	-	-	Target	Extension
TreeModel	modelName missingValueStrategy missingValuePenalty	functionName (required) algorithmName noTrueChildStrategy splitCharacteristic isScorable (only "true" is supported)	-	ModelVerification	MiningSchema Output Node	Extension ModelStats ModelExplanation Targets LocalTransformations
True	-	-	-	-	-	Extension
ScoreDistribution	-	value (required) recordCount (required)	-	-	-	Extension
Value	displayValue	value (required) property	-	-	-	Extension