← All types
Ob Object

TabularModelSpec

Tabular model configuration (regression/classification)

Fields

Target column for prediction

Feature columns used for training

Ordered data-pipeline stages (empty = source-only)

XGBoost hyperparameters

Rows kept back from training to score the model; null means the default random split

Referenced by (1)
Exact type SDL
"""Tabular model configuration (regression/classification)"""
type TabularModelSpec {
  """Target column for prediction"""
  targetColumn: String!
  """Feature columns used for training"""
  featureColumns: [String!]!
  """Ordered data-pipeline stages (empty = source-only)"""
  normalizationStages: [MLNormalizationStage!]!
  """XGBoost hyperparameters"""
  hyperparameters: MLHyperparameters
  """
  Rows kept back from training to score the model; null means the default random split
  """
  holdout: MLHoldoutSpec
}