PatientLevelPrediction (version 6.4.0)

pfi: Permutation Feature Importance

Description

Calculate the permutation feature importance (pfi) for a PLP model.

Usage

pfi(
  plpResult,
  population,
  plpData,
  repeats = 1,
  covariates = NULL,
  cores = NULL,
  log = NULL,
  logthreshold = "INFO"
)

Value

A dataframe with the covariateIds and the pfi (change in AUC caused by permuting the covariate) value

Arguments

plpResult

An object of type runPlp

population

The population created using createStudyPopulation() who will have their risks predicted

plpData

An object of type plpData - the patient level prediction data extracted from the CDM.

repeats

The number of times to permute each covariate

covariates

A vector of covariates to calculate the pfi for. If NULL it uses all covariates included in the model.

cores

Number of cores to use when running this (it runs in parallel)

log

A location to save the log for running pfi

logthreshold

The log threshold (e.g., INFO, TRACE, ...)

Details

The function permutes the each covariate/features repeats times and calculates the mean AUC change caused by the permutation.

Examples

  # takes too long 
library(dplyr)
# simulate some data
data("simulationProfile")
plpData <- simulatePlpData(simulationProfile, n=1000)
# now fit a model
saveLoc <- file.path(tempdir(), "pfi")
plpResult <- runPlp(plpData, outcomeId = 3, saveDirectory = saveLoc)
population <- createStudyPopulation(plpData, outcomeId = 3)
pfi(plpResult, population, plpData, repeats = 1, cores = 1)
# compare to model coefficients
plpResult$model$covariateImportance %>% dplyr::filter(.data$covariateValue != 0)
# clean up
unlink(saveLoc, recursive = TRUE)