CANONICAL TYPE REFERENCE

TrainingMetrics

Training metrics from a completed training run

rmseFloat

Root mean squared error (regression)

maeFloat

Mean absolute error (regression)

maxErrorFloat

Maximum prediction error (regression)

r2Float

R² coefficient of determination (regression)

accuracyFloat

Classification accuracy (0-1)

loglossFloat

Log loss (classification)

f1Float

F1 score (binary classification)

bestRoundInt

Round with best evaluation metric

totalRoundsInt

Total boosting rounds used

View exact SDL

"""Training metrics from a completed training run"""
type TrainingMetrics {
  """Root mean squared error (regression)"""
  rmse: Float
  """Mean absolute error (regression)"""
  mae: Float
  """Maximum prediction error (regression)"""
  maxError: Float
  """R² coefficient of determination (regression)"""
  r2: Float
  """Classification accuracy (0-1)"""
  accuracy: Float
  """Log loss (classification)"""
  logloss: Float
  """F1 score (binary classification)"""
  f1: Float
  """Round with best evaluation metric"""
  bestRound: Int
  """Total boosting rounds used"""
  totalRounds: Int
}

One stable address across all schema views. Follow a field’s type to explore connections between domains.

Appearance