predictKnn: Predict using a K-nearest neighbor (KNN) classifier

In OHDSI/BigKnn: Large Scale K-Nearest Neighbor Classifier using the Lucene Search Engine

Predict using a K-nearest neighbor (KNN) classifier

Description

predictKnn uses a KNN classifier to generate predictions.

Usage

predictKnn(
  cohorts,
  covariates,
  indexFolder,
  k = 1000,
  weighted = TRUE,
  threads = 1
)

Arguments

cohorts	An Andromeda table containing the cohorts with predefined columns (see below).
covariates	An Andromeda table containing the covariates with predefined columns (see below).
indexFolder	Path to a local folder where the KNN classifier index can be stored.
k	The number of nearest neighbors to use to predict the outcome.
weighted	Should the prediction be weighted by the (inverse of the ) distance metric?
threads	Number of parallel threads to used for the computation.

Details

These columns are expected in the covariates object:

⁠rowId⁠	(integer)	Row ID is used to link multiple covariates (x) to a single outcome (y)
⁠covariateId⁠	(integer)	A numeric identifier of a covariate
⁠covariateValue⁠	(real)	The value of the specified covariate

This column is expected in the covariates object:

⁠rowId⁠	(integer)	Row ID is used to link multiple covariates (x) to a single outcome (y)

Value

A data.frame with two columns:

⁠rowId⁠	(integer)	Row ID is used to link multiple covariates (x) to a single outcome (y)
⁠prediction⁠	(real)	A number between 0 and 1 representing the probability of the outcome

OHDSI/BigKnn documentation built on March 21, 2023, 8:35 a.m.