piRF (version 0.1.0)

CQRF: implements RF prediction interval using split conformal prediction as outlined in Romano, Patterson, Candes 2018. Helper function.

Description

This function implements split conformal prediction intervals for RFs. Currently used in rfint().

Usage

CQRF(
  formula = NULL,
  train_data = NULL,
  pred_data = NULL,
  num_trees = NULL,
  min_node_size = NULL,
  m_try = NULL,
  keep_inbag = TRUE,
  intervals = TRUE,
  alpha = NULL,
  forest_type = "RF",
  num_threads = NULL,
  interval_type = NULL
)

Arguments

formula

Object of class formula or character describing the model to fit. Interaction terms supported only for numerical variables.

train_data

Training data of class data.frame, matrix, dgCMatrix (Matrix) or gwaa.data (GenABEL). Matches ranger() requirements.

pred_data

Test data of class data.frame, matrix, dgCMatrix (Matrix) or gwaa.data (GenABEL). Utilizes ranger::predict() to get prediction intervals for test data.

num_trees

Number of trees.

min_node_size

Minimum number of observations before split at a node.

m_try

Number of variables to randomly select from at each split.

keep_inbag

Saves matrix of observations and which tree(s) they occur in. Required to be true to generate variance estimates for Ghosal, Hooker 2018 method. *Should not be an option...

intervals

Generate prediction intervals or not.

alpha

Significance level for prediction intervals.

forest_type

Determines what type of forest: regression forest vs. quantile regression forest. *Should not be an option...

num_threads

The number of threads to use in parallel. Default is the current number of cores.

interval_type

Type of prediction interval to generate. Options are method = c("two-sided", "lower", "upper"). Default is method = "two-sided".