Skip to contents

Extract the top N predictors per model

Source: R/PredictionQueries.R
getPredictionTopPredictors.Rd

This function extracts the top N predictors per model from the prediction results tables

Usage

getPredictionTopPredictors(
  connectionHandler,
  schema,
  plpTablePrefix = "plp_",
  cgTablePrefix = "cg_",
  targetIds = NULL,
  outcomeIds = NULL,
  numberPredictors = 100
)

Arguments

connectionHandler

A connection handler that connects to the database and extracts sql queries. Create a connection handler via `ResultModelManager::ConnectionHandler$new()`.

schema

The result database schema (e.g., 'main' for sqlite)

plpTablePrefix

The prefix used for the patient level prediction results tables

cgTablePrefix

The prefix used for the cohort generator results tables

targetIds

A vector of integers corresponding to the target cohort IDs

outcomeIds

A vector of integers corresponding to the outcome cohort IDs

numberPredictors

the number of predictors per model to return

Value

Returns a data.frame with the columns:

  • databaseName the name of the database the model was developed on

  • tarStartDay the time-at-risk start day

  • tarStartAnchor whether the time-at-risk start is relative to cohort start or end

  • tarEndDay the time-at-risk end day

  • tarEndAnchor whether the time-at-risk end is relative to cohort start or end

  • performanceId a unique identifier for the performance

  • covariateId the FeatureExtraction covariate identifier

  • covariateName the name of the covariate

  • conceptId the covariates corresponding concept or 0

  • covariateValue the feature importance or coefficient value

  • covariateCount how many people had the covariate

  • covariateMean the fraction of the target population with the covariate

  • covariateStDev the standard deviation

  • withNoOutcomeCovariateCount the number of the target population without the outcome with the covariate

  • withNoOutcomeCovariateMean the fraction of the target population without the outcome with the covariate

  • withNoOutcomeCovariateStDev the covariate standard deviation of the target population without the outcome

  • withOutcomeCovariateCount the number of the target population with the outcome with the covariate

  • withOutcomeCovariateMean the fraction of the target population with the outcome with the covariate

  • withOutcomeCovariateStDev the covariate standard deviation of the target population with the outcome

  • standardizedMeanDiff the standardized mean difference comparing the target population with outcome and without the outcome

  • rn the row number showing the covariate rank

Details

Specify the connectionHandler, the resultDatabaseSettings and (optionally) any targetIds or outcomeIds to restrict models to

See also

Other Prediction: getPredictionCohorts(), getPredictionDiagnosticTable(), getPredictionDiagnostics(), getPredictionHyperParamSearch(), getPredictionIntercept(), getPredictionModelDesigns(), getPredictionPerformanceTable(), getPredictionPerformances()

Examples

conDet <- getExampleConnectionDetails()

connectionHandler <- ResultModelManager::ConnectionHandler$new(conDet)
#> Connecting using SQLite driver

topPreds <- getPredictionTopPredictors(
  connectionHandler = connectionHandler, 
  schema = 'main',
  targetIds = 1,
  outcomeIds = 3
)