Predict new data with a trained model — predict_rROC

A convenience function to predict new data with a trained model. The function is a wrapper around the h2o.predict function. It returns a list with the predictions and optionally the metrics.

Usage

predict_rROC_h2o(
  h2o_model,
  x_prepared,
  y,
  init_h2o = FALSE,
  calculate_metrics = TRUE,
  sample_split = NULL
)

Arguments

h2o_model: The model to use for prediction.
x_prepared: A data frame containing the preprocessed data to use in the model. The data frame can contain more variables than necessary within h2o_model
y: The true outcome variable. Usually a factor.
init_h2o: Whether to initialize h2o. Defaults to FALSE as I expect it was initialized during training already.
calculate_metrics: Whether to calculate metrics. Defaults to TRUE.
sample_split: A vector of length(y) containing the "split" of the data. The metrices are calculated for each split separately. Defaults to NULL, which means that the whole data is used.