int_conformal_split: Prediction intervals via split conformal inference In probably: Tools for Post-Processing Predicted Values

 int_conformal_split R Documentation

Prediction intervals via split conformal inference

Description

Nonparametric prediction intervals can be computed for fitted regression workflow objects using the split conformal inference method described by Lei et al (2018).

Usage

``````int_conformal_split(object, ...)

## Default S3 method:
int_conformal_split(object, ...)

## S3 method for class 'workflow'
int_conformal_split(object, cal_data, ...)
``````

Arguments

 `object` A fitted `workflows::workflow()` object. `...` Not currently used. `cal_data` A data frame with the original predictor and outcome data used to produce predictions (and residuals). If the workflow used a recipe, this should be the data that were inputs to the recipe (and not the product of a recipe).

Details

This function implements what is usually called "split conformal inference" (see Algorithm 1 in Lei et al (2018)).

This function prepares the statistics for the interval computations. The `predict()` method computes the intervals for new data and the signficance level is specified there.

`cal_data` should be large enough to get a good estimates of a extreme quantile (e.g., the 95th for 95% interval) and should not include rows that were in the original training set.

Value

An object of class `"int_conformal_split"` containing the information to create intervals (which includes `object`). The `predict()` method is used to produce the intervals.

References

Lei, Jing, et al. "Distribution-free predictive inference for regression." Journal of the American Statistical Association 113.523 (2018): 1094-1111.

`predict.int_conformal_split()`

Examples

``````
library(workflows)
library(dplyr)
library(parsnip)
library(rsample)
library(tune)
library(modeldata)

set.seed(2)
sim_train <- sim_regression(500)
sim_cal <- sim_regression(200)
sim_new <- sim_regression(5) %>% select(-outcome)

# We'll use a neural network model
mlp_spec <-
mlp(hidden_units = 5, penalty = 0.01) %>%
set_mode("regression")

mlp_wflow <-
workflow() %>%