conform.classification: Generate a conform.classification object

In labuzzetta/conformr: Conformal Prediction Research R Package

Generate a conform.classification object

Description

Generate a conform.classification object

Usage

conform.classification(
  x,
  y,
  algorithm = c("standard", "mondrian", "knn", "weighted", "double"),
  type = c("ratio", "diff", "vovk"),
  train_method = "glm",
  train_p = 0.5,
  train_n = NULL,
  cal_n = NULL,
  cal_id = NULL,
  train_id = NULL,
  pretrained = NULL,
  ...
)

Arguments

x	data.frame or matrix of covariate data
y	vector of categorical responses
algorithm	method of conformal prediction
type	nonconformity score type
train_method	machine learning method from caret package for underlying model
train_p	size of proper training dataset by proportion of observations in x and y (default 0.5 split if train_n or train_id not specified)
train_n	size of proper training dataset by number n observations from x and y (both cal_n and train_n must be specified for this option)
cal_n	size of calibration dataset by number n observations from x and y (both cal_n and train_n must be specified for this option)
cal_id	list of indexes of selected calibration observations
train_id	list of indexes of selected proper training set observations (overides train_n and train_p)
pretrained	underlying caret model pretrained to pass into function
...	additional parameters for caret model training

Value

conform.classification object

labuzzetta/conformr documentation built on May 13, 2022, 12:50 a.m.