config

Type Members

1. case class AlgorithmConfig(stringBoundaries: Map[String, List[String]], numericBoundaries: Map[String, (Double, Double)]) extends Product with Serializable

   

2. trait BatteryDefaults extends AnyRef

   

3. class BatteryGenerator extends BatteryDefaults with ConfigurationDefaults

   

   A class and companion object to generate configurations for a particular modeling type.

   A class and companion object to generate configurations for a particular modeling type. Developer API. Currently available modeling types: regressor and classifier

   Example:

   1. val classifierConfigs = new BatterryGenerator("regressor").setModelsToTest(Array("RandomForest", "GBT")) .generateDefaultConfigs( "labelColumn", "features", "https://instance.host.com", "regressorTest", "<mytoken>", "/model/save/dir", "/inference/save/dir", "full", "_best_model", true, false)

   Since

   0.5.0.3

4. trait ConfigurationDefaults extends AnyRef

   

5. class ConfigurationGenerator extends ConfigurationDefaults

   

   Main Configuration Generator utility class, used for generating a modeling configuration to execute the autoML framework.

   Main Configuration Generator utility class, used for generating a modeling configuration to execute the autoML framework.

   Since

   0.5

6. case class FeatureEngineeringConfig(dataPrepParallelism: Int, numericFillStat: String, characterFillStat: String, modelSelectionDistinctThreshold: Int, outlierFilterBounds: String, outlierLowerFilterNTile: Double, outlierUpperFilterNTile: Double, outlierFilterPrecision: Double, outlierContinuousDataThreshold: Int, outlierFieldsToIgnore: Array[String], pearsonFilterStatistic: String, pearsonFilterDirection: String, pearsonFilterManualValue: Double, pearsonFilterMode: String, pearsonAutoFilterNTile: Double, covarianceCorrelationCutoffLow: Double, covarianceCorrelationCutoffHigh: Double, scalingType: String, scalingMin: Double, scalingMax: Double, scalingStandardMeanFlag: Boolean, scalingStdDevFlag: Boolean, scalingPNorm: Double, featureImportanceCutoffType: String, featureImportanceCutoffValue: Double, dataReductionFactor: Double, cardinalitySwitch: Boolean, cardinalityType: String, cardinalityLimit: Int, cardinalityPrecision: Double, cardinalityCheckMode: String, filterPrecision: Double, categoricalNAFillMap: Map[String, String], numericNAFillMap: Map[String, AnyVal], characterNABlanketFillValue: String, numericNABlanketFillValue: Double, naFillMode: String, featureInteractionRetentionMode: String, featureInteractionContinuousDiscretizerBucketCount: Int, featureInteractionParallelism: Int, featureInteractionTargetInteractionPercentage: Double) extends Product with Serializable

   

7. case class GenericConfig(labelCol: String, featuresCol: String, dateTimeConversionType: String, fieldsToIgnoreInVector: Array[String], scoringMetric: String, scoringOptimizationStrategy: String) extends Product with Serializable

   

8. class GenericConfigGenerator extends ConfigurationDefaults

   

9. case class InstanceConfig(modelFamily: String, predictionType: String, genericConfig: GenericConfig, switchConfig: SwitchConfig, featureEngineeringConfig: FeatureEngineeringConfig, algorithmConfig: AlgorithmConfig, tunerConfig: TunerConfig, loggingConfig: LoggingConfig) extends Product with Serializable

   

10. class InstanceConfigValidation extends ConfigurationDefaults

    

11. case class LoggingConfig(mlFlowLoggingFlag: Boolean, mlFlowLogArtifactsFlag: Boolean, mlFlowTrackingURI: String, mlFlowExperimentName: String, mlFlowAPIToken: String, mlFlowModelSaveDirectory: String, mlFlowLoggingMode: String, mlFlowBestSuffix: String, inferenceConfigSaveLocation: String, mlFlowCustomRunTags: Map[String, String]) extends Product with Serializable

    

12. case class SwitchConfig(naFillFlag: Boolean, varianceFilterFlag: Boolean, outlierFilterFlag: Boolean, pearsonFilterFlag: Boolean, covarianceFilterFlag: Boolean, oneHotEncodeFlag: Boolean, scalingFlag: Boolean, featureInteractionFlag: Boolean, dataPrepCachingFlag: Boolean, autoStoppingFlag: Boolean, pipelineDebugFlag: Boolean) extends Product with Serializable

    

13. case class TunerConfig(tunerAutoStoppingScore: Double, tunerParallelism: Int, tunerKFold: Int, tunerTrainPortion: Double, tunerTrainSplitMethod: String, tunerKSampleSyntheticCol: String, tunerKSampleKGroups: Int, tunerKSampleKMeansMaxIter: Int, tunerKSampleKMeansTolerance: Double, tunerKSampleKMeansDistanceMeasurement: String, tunerKSampleKMeansSeed: Long, tunerKSampleKMeansPredictionCol: String, tunerKSampleLSHHashTables: Int, tunerKSampleLSHSeed: Long, tunerKSampleLSHOutputCol: String, tunerKSampleQuorumCount: Int, tunerKSampleMinimumVectorCountToMutate: Int, tunerKSampleVectorMutationMethod: String, tunerKSampleMutationMode: String, tunerKSampleMutationValue: Double, tunerKSampleLabelBalanceMode: String, tunerKSampleCardinalityThreshold: Int, tunerKSampleNumericRatio: Double, tunerKSampleNumericTarget: Int, tunerTrainSplitChronologicalColumn: String, tunerTrainSplitChronologicalRandomPercentage: Double, tunerSeed: Long, tunerFirstGenerationGenePool: Int, tunerNumberOfGenerations: Int, tunerNumberOfParentsToRetain: Int, tunerNumberOfMutationsPerGeneration: Int, tunerGeneticMixing: Double, tunerGenerationalMutationStrategy: String, tunerFixedMutationValue: Int, tunerMutationMagnitudeMode: String, tunerEvolutionStrategy: String, tunerGeneticMBORegressorType: String, tunerGeneticMBOCandidateFactor: Int, tunerContinuousEvolutionImprovementThreshold: Int, tunerContinuousEvolutionMaxIterations: Int, tunerContinuousEvolutionStoppingScore: Double, tunerContinuousEvolutionParallelism: Int, tunerContinuousEvolutionMutationAggressiveness: Int, tunerContinuousEvolutionGeneticMixing: Double, tunerContinuousEvolutionRollingImprovingCount: Int, tunerModelSeed: Map[String, Any], tunerHyperSpaceInference: Boolean, tunerHyperSpaceInferenceCount: Int, tunerHyperSpaceModelCount: Int, tunerHyperSpaceModelType: String, tunerInitialGenerationMode: String, tunerInitialGenerationPermutationCount: Int, tunerInitialGenerationIndexMixingMode: String, tunerInitialGenerationArraySeed: Long, tunerOutputDfRepartitionScaleFactor: Int, tunerDeltaCacheBackingDirectory: String, tunerDeltaCacheBackingDirectoryRemovalFlag: Boolean, splitCachingStrategy: String) extends Product with Serializable