Predict from a fitted KAF model

Usage

# S3 method for class 'kaf_fit'
predict(
  object,
  newdata,
  type = c("response", "prob", "class", "link"),
  threshold = 0.5,
  as_tensor = FALSE,
  ...
)

Arguments

object: A fitted object returned by kaf_fit().
newdata: Matrix, data frame, vector, or 2D torch tensor.
type: Character. Prediction type. "response" returns regression predictions for regression, probabilities for classification. "prob" returns probabilities for classification. "class" returns predicted classes. "link" returns raw model logits/outputs.
threshold: Numeric. Classification threshold used for binary class predictions.
as_tensor: Logical. If TRUE, return a torch tensor where supported.
...: Unused.

Value

Predictions as a vector, matrix, factor, or torch tensor.